Dissertations / Theses: 'Knowledge Graphs'

1

Ghiasnezhad, Omran Pouya. "Rule Learning in Knowledge Graphs." Thesis, Griffith University, 2018. http://hdl.handle.net/10072/382680.

Full text

Abstract:

With recent advancements in knowledge extraction and knowledge management systems, an enormous number of knowledge bases have been constructed, such as YAGO, and Wikidata. These automatically built knowledge bases which contain millions of entities and their relations have been stored in graph-based schemas, and thus are usually referred to as knowledge graphs (KGs). Since KGs have been built based on the limited available data, they are far from complete. However, learning frequent patterns in the form of logical rules from these incomplete KGs has two main advantages. First, by applying the learned rules, we can infer new facts, so we could complete the KGs. Second, the rules are stand-alone knowledge which express valuable insight about the data. However, learning rules from KGs in relation to the real-world scenarios imposes several challenges. First, due to the vast size of real-world KGs, developing a rule learning method is challenging. In fact, existing methods are not scalable for learning rst order rules, while various optimisation strategies are used such as sampling and language bias (i.e., restrictions on the form of rules). Second, applying the learned rules to the vast KG and inferring new facts is another di cult issue. Learned rules usually contain a lot of noises and adding new facts can cause inconsistency of KGs. Third, it is useful but non-trivial to extend an existing method of rule learning to the case of stream KGs. Forth, in many data repositories, the facts are augmented with time stamps. In this case, we face a stream of data (KGs). Considering time as a new dimension of data imposes some challenges to the rule learning process. It would be useful to construct a time-sensitive model from the stream of data and apply the obtained model to stream KGs. Last, the density of information in a KG is varied. Although the size of a KG is vast, it contains a limited amount of information for some relations. Consequently, that part of KG is sparse. Learning a set of accurate and informative rules regarding the sparse part of a KG is challenging due to the lack of su cient training data. In this thesis, we investigate these research problems and present our methods for rule learning in various scenarios. We have rst developed a new approach, named Rule Learning via Learning Representation (RLvLR), to learning rules from KGs by using the technique of embedding in representation learning together with a new sampling method. RLvLR learns rst-order rules from vast KGs by exploring the embedding space. It can handle some large KGs that cannot be handled by existing rule learners e ciently, due to a novel sampling method. To improve the performance of RLvLR for handling sparse data, we propose a transfer learning method, Transfer Rule Learner (TRL), for rule learning. Based on a similarity characterised by the embedding representation, our method is able to select most relevant KGs and rules to transfer from a pool of KGs whose rules have been obtained. We have also adapted RLvLR to handle stream KGs instead of static KGs. Then a system called StreamLearner is developed for learning rules from stream KGs. These proposed methods can only learn so-called closed path rules, which is a proper subset of Horn rules. Thus, we have also developed a transfer rule learner (T-LPAD) that learns the structure of logic program with annotated disjunctions. T-LPAD is created by employing transfer learning to explore the space of rules' structures more e ciently. Various experiments have been conducted to test and validate the proposed methods. Our experimental results show that our methods outperform state-of-the-art methods in many ways.
Thesis (PhD Doctorate)
Doctor of Philosophy (PhD)
School of Info & Comm Tech
Science, Environment, Engineering and Technology
Full Text

APA, Harvard, Vancouver, ISO, and other styles

2

Hossayni, Hicham. "Enabling industrial maintenance knowledge sharing by using knowledge graphs." Electronic Thesis or Diss., Institut polytechnique de Paris, 2021. http://www.theses.fr/2021IPPAS017.

Full text

Abstract:

Autrefois considérée comme faisant partie des coûts généraux de l'entreprise, la maintenance industrielle est devenue critique pour la continuité de l'activité et une véritable source de données. Malgré les sommes importantes investies par les entreprises dans la fabrication intelligente, les pratiques traditionnelles en maintenance dominent toujours le paysage industriel. Dans cette thèse, nous étudions le partage des connaissances comme une solution potentielle qui peut inverser la tendance et améliorer l'activité de maintenance pour se conformer à l'esprit de l'industrie 4.0. Nous considérons spécifiquement les graphes de connaissances comme un outil permettant de partager les connaissances de maintenance entre les différents acteurs de l'industrie.Dans la première contribution de cette thèse, nous avons mené une étude de terrain à travers une campagne d'entretiens avec des experts aux profils différents et issus de divers domaines industriels. Cela nous a permis de tester l'hypothèse de l'amélioration de l'activité de maintenance via le partage des connaissances, qui est un concept assez nouveau dans de nombreuses industries. Les résultats de cette activité montrent clairement un intérêt réel pour notre approche et révèlent les différents besoins et défis à relever.La deuxième contribution est la conception et le prototype de "SemKoRe"; une solution s'appuyant sur le Web sémantique pour partager les connaissances de maintenance. Elle collecte les données liées aux défaillances de machine, les structure dans un graphe de connaissances et les partage entre tous les clients connectés pour réparer facilement les futures défaillances du même type. Une architecture flexible a été proposée pour couvrir les besoins des différents clients. SemKoRe a reçu l'approbation de plusieurs clients de Schneider implantés dans plusieurs pays et de différents segments.Dans la troisième contribution, nous avons conçu et mis en oeuvre une nouvelle solution pour la détection automatique des données sensibles dans les rapports de maintenance. En effet, les rapports de maintenance peuvent contenir des données confidentielles susceptibles de compromettre ou d'avoir un impact négatif sur l'activité de l'entreprise si elles sont révélées. Cette fonctionnalité est perçue, par les experts du domaine comme un point essentiel et critique pour SemKoRe. Elle permet d'éviter la divulgation de données sensibles lors de l'activité de partage des connaissances. Dans cette contribution, nous nous sommes appuyés sur le web sémantique et le traitement du langage naturel pour développer des modèles personnalisés pour la détection de données sensibles. La construction et l'apprentissage de tels modèles nécessitent une quantité considérable de données. Par conséquent, nous avons mis en place plusieurs services pour la collecte collaborative de données, l'annotation de texte et la construction de corpus. Aussi, une architecture et un workflow simplifié ont été proposés pour la génération et le déploiement de modèles de détection de données sensibles personnalisables sur les passerelles de périphérie.En plus de ces contributions, nous avons travaillé sur différentes fonctionnalités connexes à forte valeur ajoutée pour le projet SemKoRe, et qui ont abouti à différents brevets. Par exemple, nous avons développé et breveté une nouvelle méthode pour interagir avec les données de séries chronologiques à l'aide de critères sémantiques. Elle combine l'utilisation d'ontologies et de bases de données de séries chronologiques pour offrir un ensemble utile de fonctionnalités même sur des passerelles périphériques aux ressources limitées. Nous avons également conçu un nouvel outil qui aide les développeurs à interagir facilement avec des graphes de connaissances avec peu ou pas de connaissance des technologies du Web sémantique. Cette solution a été brevetée et s'avère utile pour d'autres projets basés sur des ontologies
Formerly considered as part of general enterprise costs, industrial maintenance has become critical for business continuity and a real source of data. Despite the heavy investments made by companies in smart manufacturing, traditional maintenance practices still dominate the industrial landscape. In this Ph.D., we investigate maintenance knowledge sharing as a potential solution that can invert the trend and enhance the maintenance activity to comply with the Industry 4.0 spirit. We specifically consider the knowledge graphs as an enabler to share the maintenance knowledge among the different industry players.In the first contribution of this thesis, we conducted a field study through a campaign of interviews with different experts with different profiles and from different industry domains. This allowed us to test the hypothesis of improving the maintenance activity via knowledge sharing which is quite a novel concept in many industries. The results of this activity clearly show a real interest in our approach and reveal the different requirements and challenges that need to be addressed.The second contribution is the concept, design, and prototype of "SemKoRe" which is a vendor-agnostic solution relying on Semantic Web technologies to share the maintenance knowledge. It gathers all machine failure-related data in the knowledge graph and shares it among all connected customers to easily solve future failures of the same type. A flexible architecture was proposed to cover the varied needs of the different customers. SemKoRe received approval of several Schneider clients located in several countries and from various segments.In the third contribution, we designed and implemented a novel solution for the automatic detection of sensitive data in maintenance reports. In fact, maintenance reports may contain some confidential data that can compromise or negatively impact the company's activity if revealed. This feature came up as the make or break point for SemKoRe for the interviewed domain experts. It allows avoiding sensitive data disclosure during the knowledge-sharing activity. In this contribution, we relied on semantic web and natural language processing techniques to develop custom models for sensitive data detection. The construction and training of such models require a considerable amount of data. Therefore, we implemented several services for collaborative data collection, text annotation, and corpus construction. Also, an architecture and a simplified workflow were proposed for the generation and deployment of customizable sensitive data detection models on edge gateways.In addition to these contributions, we worked on different peripheral features with a strong value for the SemKoRe project, and that has resulted in different patents. For instance, we prototyped and patented a novel method to query time series data using semantic criteria. It combines the use of ontologies and time-series databases to offer a useful set of querying capabilities even on resource-constrained edge gateways. We also designed a novel tool that helps software developers to easily interact with knowledge graphs with little or no knowledge of semantic web technologies. This solution has been patented and turns out to be useful for other ontology-based projects

APA, Harvard, Vancouver, ISO, and other styles

3

Xu, Keyulu. "Graph structures, random walks, and all that : learning graphs with jumping knowledge networks." Thesis, Massachusetts Institute of Technology, 2019. https://hdl.handle.net/1721.1/121660.

Full text

Abstract:

This electronic version was submitted by the student author. The certified thesis is available in the Institute Archives and Special Collections.
Thesis: S.M., Massachusetts Institute of Technology, Department of Electrical Engineering and Computer Science, 2019
Cataloged from student-submitted PDF version of thesis.
Includes bibliographical references (pages 51-54).
Graph representation learning aims to extract high-level features from the graph structures and node features, in order to make predictions about the nodes and the graphs. Applications include predicting chemical properties of drugs, community detection in social networks, and modeling interactions in physical systems. Recent deep learning approaches for graph representation learning, namely Graph Neural Networks (GNNs), follow a neighborhood aggregation procedure, where the representation vector of a node is computed by recursively aggregating and transforming feature vectors of its neighboring nodes. We analyze some important properties of these models, and propose a strategy to overcome the limitations. In particular, the range of neighboring nodes that a node's representation draws from strongly depends on the graph structure, analogous to the spread of a random walk. To adapt to local neighborhood properties and tasks, we explore an architecture - jumping knowledge (JK) networks that flexibly leverages, for each node, different neighborhood ranges to enable better structure-aware representation. In a number of experiments on social, bioinformatics and citation networks, we demonstrate that our model achieves state-of-the-art performance. Furthermore, combining the JK framework with models like Graph Convolutional Networks, GraphSAGE and Graph Attention Networks consistently improves those models' performance.
by Keyulu Xu.
S.M.
S.M. Massachusetts Institute of Technology, Department of Electrical Engineering and Computer Science

APA, Harvard, Vancouver, ISO, and other styles

4

Mulder, Jan A. "Using discrimination graphs to represent visual knowledge." Thesis, University of British Columbia, 1985. http://hdl.handle.net/2429/25943.

Full text

Abstract:

This dissertation is concerned with the representation of visual knowledge. Image features often have many different local interpretations. As a result, visual interpretations are often ambiguous and hypothetical. In many model-based vision systems the problem of representing ambiguous and hypothetical interpretations is not very specifically addressed. Generally, specialization hierarchies are used to suppress a potential explosion in local interpretations. Such a solution has problems, as many local interpretations cannot be represented by a single hierarchy. As well, ambiguous and hypothetical interpretations tend to be represented along more than one knowledge representation dimension limiting modularity in representation and control. In this dissertation a better solution is proposed. Classes of objects which have local features with similar appearance in the image are represented by discrimination graphs. Such graphs are directed and acyclic. Their leaves represent classes of elementary objects. All other nodes represent abstract (and sometimes unnatural) classes of objects, which intensionally represent the set of elementary object classes that descend from them. Rather than interpreting each image feature as an elementary object, we use the abstract class that represents the complete set of possible (elementary) objects. Following the principle of least commitment, the interpretation of each image feature is repeatedly forced into more restrictive classes as the context for the image feature is expanded, until the image no longer provides subclassification information. This approach is called discrimination vision, and it has several attractive features. First, hypothetical and ambiguous interpretations can be represented along one knowledge representation dimension. Second, the number of hypotheses represented for a single image feature can be kept small. Third, in an interpretation graph competing hypotheses can be represented in the domain of a single variable. This often eliminates the need for restructuring the graph when a hypothesis is invalidated. Fourth, the problem of resolving ambiguity can be treated as a constraint satisfaction problem which is a well researched problem in Computational Vision. Our system has been implemented as Mapsee-3, a program for interpreting sketch maps. A hierarchical arc consistency algorithm has been used to deal with the inherently hierarchical discrimination graphs. Experimental data show that, for the domain implemented, this algorithm is more efficient than standard arc consistency algorithms.
Science, Faculty of
Computer Science, Department of
Graduate

APA, Harvard, Vancouver, ISO, and other styles

5

Sandelius, Hugo. "Creating Knowledge Graphs using Distributional Semantic Models." Thesis, KTH, Skolan för datavetenskap och kommunikation (CSC), 2016. http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-199702.

Full text

Abstract:

This report researches a method for creating knowledge graphs, a specific way of structuring information, using distributional semantic models. Two different algorithms for selecting graph edges and two different algorithms for labelling edges are tried, and variations of those are evaluated. We perform experiments comparing our knowledge graphs with existing manually constructed knowledge graphs of high quality, with respect to graph structure and edge labels. We find that the algorithms usually produces graphs with a structure similar to that of manually constructed knowledge graphs, as long as the data set is sufficiently large and general, and that the similarity of edge labels to manually chosen edge labels vary widely depending on input.

APA, Harvard, Vancouver, ISO, and other styles

6

Salehpour, Masoud. "High-performance Query Processing over Knowledge Graphs." Thesis, The University of Sydney, 2022. https://hdl.handle.net/2123/28569.

Full text

Abstract:

The label “Knowledge Graph” (KG) has been used in the literature for over four decades, typically to refer to a collection of information about real-world entities and their inter-relationships. The proliferation of KGs in recent times opens up exciting opportunities for a broad range of semantic applications such as recommendations. However, unlocking the full potential of KGs in response to the growing deployment requires data platforms to efficiently store and process the content to support various applications. What began with extensions of relational database systems to store the content of KGs led to the design and development of a number of new specialized data management systems. Although progress has been made around building efficient KG data management systems, developing high-performance systems continues to pose research challenges. In this research, we studied the efficiency of existing systems for storing and processing KG content. Our results pointed to performance inconsistencies in representative systems across diverse query types. We address this by introducing a polyglot model of KG query processing to analyze each query and match it to the best-performing available systems. Experimental evaluation highlighted that our proposed approach provides consistently high performance. Finally, we investigated leveraging emerging hardware and its benefits to RDF data management and performance. To this end, we introduced a novel index structure, RDFix, that utilizes Persistent Memory (PM) to outperform existing read-optimized indexes as shown experimentally.

APA, Harvard, Vancouver, ISO, and other styles

7

BONOMO, Mariella. "Knowledge Extraction from Biological and Social Graphs." Doctoral thesis, Università degli Studi di Palermo, 2022. https://hdl.handle.net/10447/576508.

Full text

Abstract:

Many problems from real life deal with the generation of enormous, varied, dynamic, and interconnected datasets coming from different and heterogeneous sources. Analysing large volumes of data makes it possible to generate new knowledge useful for making more informed decisions, in business and beyond. From personalising customer communication to streamlining production processes, via flow and emergency management, Big Data Analytics has an impact on all processes. The potential uses of Big Data go much further: two of the largest sources of data are including individual traders’ purchasing history, the use of Biological Networks for disease prediction or the reduction and study of Biological Networks. From a computer science point of view, the networks are graphs with various characteristics specific to the application domain. This PhD Thesis focuses on the proposal of novel knowledge extraction techniques from large graphs, mainly based on Big Data methodologies. Two application contexts are considered and three specific problems have been solved: Social data, for the optimization of advertising campaigns, the comparison of user profiles, and neighborhood analysis. Biological and Medical data, with the final aim of identifying biomarkers for diagnosis, treatment, prognosis, and prevention of diseases.

APA, Harvard, Vancouver, ISO, and other styles

8

Simonne, Lucas. "Mining differential causal rules in knowledge graphs." Electronic Thesis or Diss., université Paris-Saclay, 2023. http://www.theses.fr/2023UPASG008.

Full text

Abstract:

La fouille de règles d'association au sein de graphes de connaissances est un domaine de recherche important.En effet, ce type de règle permet de représenter des connaissances, et leur application permet de compléter un graphe en ajoutant des données manquantes ou de supprimer des données erronées.Cependant, ces règles ne permettent pas d'exprimer des relations causales, dont la sémantique diffère d'une simple association ou corrélation. Dans un système, un lien de causalité entre une variable A et une variable B est une relation orientée de A vers B et indique qu'un changement dans A cause un changement dans B, les autres variables du système conservant les mêmes valeurs.Plusieurs cadres d'étude existent pour déterminer des relations causales, dont le modèle d'étude des résultats potentiels, qui consiste à apparier des instances similaires ayant des valeurs différentes sur une variable nommée traitement pour étudier l'effet de ce traitement sur une autre variable nommée résultat.Nous proposons dans cette thèse plusieurs approches permettant de définir des règles représentant l'effet causal d'un traitement sur un résultat.Cet effet peut être local, i.e., valide pour un sous-ensemble d'instances d'un graphe de connaissances défini par un motif de graphe, ou bien moyen, i.e., valide en moyenne pour l'ensemble d'instances de la classe considérée. La découverte de ces règles se base sur le cadre d'étude des résultats potentiels en appariant des instances similaires, en comparant leurs descriptions RDF au sein du graphe ou bien leurs représentations vectorielles apprises à travers des modèles de plongements de graphes
The mining of association rules within knowledge graphs is an important area of research.Indeed, this type of rule makes it possible to represent knowledge, and their application makes it possible to complete a knowledge graph by adding missing triples or to remove erroneous triples.However, these rules express associations and do not allow the expression of causal relations, whose semantics differ from an association or a correlation.In a system, a causal link between variable A and variable B is a relationship oriented from A to B. It indicates that a change in A causes a change in B, with the other variables in the system maintaining the same values.Several frameworks exist for determining causal relationships, including the potential outcome framework, which involves matching similar instances with different values on a variable named treatment to study the effect of that treatment on another variable named the outcome.In this thesis, we propose several approaches to define rules representing a causal effect of a treatment on an outcome.This effect can be local, i.e., valid for a subset of instances of a knowledge graph defined by a graph pattern, or average, i.e., valid on average for the whole set of graph instances.The discovery of these rules is based on the framework of studying potential outcomes by matching similar instances and comparing their RDF descriptions or their learned vectorial representations through graph embedding models

APA, Harvard, Vancouver, ISO, and other styles

9

Boschin, Armand. "Machine learning techniques for automatic knowledge graph completion." Electronic Thesis or Diss., Institut polytechnique de Paris, 2023. http://www.theses.fr/2023IPPAT016.

Full text

Abstract:

Un graphe de connaissances est un graphe orienté dont les nœuds sont des entités et les arêtes, typées par une relation, représentent des faits connus liant les entités. Ces graphes sont capables d'encoder une grande variété d'information mais leur construction et leur exploitation peut se révéler complexe. Historiquement, des méthodes symboliques ont permis d'extraire des règles d'interaction entre entités et relations, afin de corriger des anomalies ou de prédire des faits manquants. Plus récemment, des méthodes d'apprentissage de représentations vectorielles, ou plongements, ont tenté de résoudre ces mêmes tâches. Initialement purement algébriques ou géométriques, ces méthodes se sont complexifiées avec les réseaux de neurones profonds et ont parfois été combinées à des techniques symboliques antérieures.Dans cette thèse, on s'intéresse tout d'abord au problème de l'implémentation. En effet, la grande diversité des bibliothèques utilisées rend difficile la comparaison des résultats obtenus par différents modèles. Dans ce contexte, la bibliothèque Python TorchKGE a été développée afin de proposer un environnement unique pour l'implémentation de modèles de plongement et un module hautement efficace d'évaluation par prédiction de liens. Cette bibliothèque repose sur l'accélération graphique de calculs tensoriels proposée par PyTorch, est compatible avec les bibliothèques d'optimisation usuelles et est disponible en source ouverte.Ensuite, les travaux portent sur l'enrichissement automatique de Wikidata par typage des hyperliens liant les articles de Wikipedia. Une étude préliminaire a montré que le graphe des articles de Wikipedia est beaucoup plus dense que le graphe de connaissances correspondant dans Wikidata. Une nouvelle méthode d'entrainement impliquant les relations et une méthode d'inférence utilisant les types des entités ont été proposées et des expériences ont montré la pertinence de l'approche, y compris sur un nouveau jeu de données.Enfin, le typage automatique d'entités est exploré comme une tâche de classification hiérarchique. Ceci a mené à la conception d'une fonction d'erreur hiérarchique, utilisée pour l'entrainement de modèles tensoriels, ainsi qu'un nouveau type d'encodeur. Des expériences ont permis une bonne compréhension de l'impact que peut avoir une connaissance a priori de la taxonomie des classes sur la classification. Elles ont aussi renforcé l'intuition que la hiérarchie peut être apprise à partir des données si le jeu est suffisamment riche
A knowledge graph is a directed graph in which nodes are entities and edges, typed by a relation, represent known facts linking two entities. These graphs can encode a wide variety of information, but their construction and exploitation can be complex. Historically, symbolic methods have been used to extract rules about entities and relations, to correct anomalies or to predict missing facts. More recently, techniques of representation learning, or embeddings, have attempted to solve these same tasks. Initially purely algebraic or geometric, these methods have become more complex with deep neural networks and have sometimes been combined with pre-existing symbolic techniques.In this thesis, we first focus on the problem of implementation. Indeed, the diversity of libraries used makes the comparison of results obtained by different models a complex task. In this context, the Python library TorchKGE was developed to provide a unique setup for the implementation of embedding models and a highly efficient inference evaluation module. This library relies on graphic acceleration of tensor computation provided by PyTorch, is compatible with widespread optimization libraries and is available as open source.We then consider the automatic enrichment of Wikidata by typing the hyperlinks linking Wikipedia pages. A preliminary study showed that the graph of Wikipedia articles is much denser than the corresponding knowledge graph in Wikidata. A new training method involving relations and an inference method using entity types were proposed and experiments showed the relevance of the combined approach, including on a new dataset.Finally, we explore automatic entity typing as a hierarchical classification task. That led to the design of a new hierarchical loss used to train tensor-based models along with a new type of encoder. Experiments on two datasets have allowed a good understanding of the impact a prior knowledge of class taxonomy can have on a classifier but also reinforced the intuition that the hierarchy can be learned from the features if the dataset is large enough

APA, Harvard, Vancouver, ISO, and other styles

10

Maher, Peter E. "A Prolog implementation of conceptual graphs." Thesis, Cardiff University, 1987. http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.257650.

Full text

APA, Harvard, Vancouver, ISO, and other styles

11

Oshurko, Ievgeniia. "Knowledge representation and curation in hierarchies of graphs." Thesis, Lyon, 2020. http://www.theses.fr/2020LYSEN024.

Full text

Abstract:

L'extraction automatique des intuitions et la construction de modèles computationnels à partir de connaissances sur des systèmes complexes repose largement sur le choix d'une représentation appropriée. Ce travail s'efforce de construire un cadre adapté pour la représentation de connaissances fragmentées sur des systèmes complexes et sa curation semi-automatisé.Un système de représentation des connaissances basé sur des hiérarchies de graphes liés à l'aide d'homomorphismes est proposé. Les graphes individuels représentent des fragments de connaissances distincts et les homomorphismes permettent de relier ces fragments. Nous nous concentrons sur la conception de mécanismes mathématiques,basés sur des approches algébriques de la réécriture de graphes, pour la transformation de graphes individuels dans des hiérarchies qui maintient des relations cohérentes entre eux.De tels mécanismes fournissent une piste d'audit transparente, ainsi qu'une infrastructure pour maintenir plusieurs versions des connaissances.La théorie développée est appliquée à la conception des schémas pour les bases de données orientée graphe qui fournissent des capacités de co-évolution schémas-données.Ensuite, cette théorie est utilisée dans la construction du cadre KAMI, qui permet la curation des connaissances sur la signalisation dans les cellules. KAMI propose des mécanismes pour une agrégation semi-automatisée de faits individuels sur les interactions protéine-protéine en corpus de connaissances, la réutilisation de ces connaissances pour l'instanciation de modèles de signalisation dans différents contextes cellulaires et la génération de modèles exécutables basés sur des règles
The task of automatically extracting insights or building computational models fromknowledge on complex systems greatly relies on the choice of appropriate representation.This work makes an effort towards building a framework suitable for representation offragmented knowledge on complex systems and its semi-automated curation---continuouscollation, integration, annotation and revision.We propose a knowledge representation system based on hierarchies of graphs relatedwith graph homomorphisms. Individual graphs situated in such hierarchies representdistinct fragments of knowledge and the homomorphisms allow relating these fragments.Their graphical structure can be used efficiently to express entities and their relations. Wefocus on the design of mathematical mechanisms, based on algebraic approaches to graphrewriting, for transformation of individual graphs in hierarchies that maintain consistentrelations between them. Such mechanisms provide a transparent audit trail, as well as aninfrastructure for maintaining multiple versions of knowledge.We describe how the developed theory can be used for building schema-aware graphdatabases that provide schema-data co-evolution capabilities. The proposed knowledgerepresentation framework is used to build the KAMI (Knowledge Aggregation and ModelInstantiation) framework for curation of cellular signalling knowledge. The frameworkallows for semi-automated aggregation of individual facts on protein-protein interactionsinto knowledge corpora, reuse of this knowledge for instantiation of signalling models indifferent cellular contexts and generation of executable rule-based models

APA, Harvard, Vancouver, ISO, and other styles

12

Gunaratna, Kalpa. "Semantics-based Summarization of Entities in Knowledge Graphs." Wright State University / OhioLINK, 2017. http://rave.ohiolink.edu/etdc/view?acc_num=wright1496124815009777.

Full text

APA, Harvard, Vancouver, ISO, and other styles

13

FUTIA, GIUSEPPE. "Neural Networks forBuilding Semantic Models and Knowledge Graphs." Doctoral thesis, Politecnico di Torino, 2020. http://hdl.handle.net/11583/2850594.

Full text

APA, Harvard, Vancouver, ISO, and other styles

14

Rawsthorne, Helen Mair. "Creation of geospatial knowledge graphs from heterogeneous sources." Electronic Thesis or Diss., Université Gustave Eiffel, 2024. http://www.theses.fr/2024UEFL2006.

Full text

Abstract:

Certaines connaissances spatiales, actuelles ou historiques, n'existent que sous forme de texte. Les guides de voyage, les documents historiques et les publications sur les réseaux sociaux sont quelques exemples de sources de connaissances spatiales non structurées. Les sources textuelles contiennent des connaissances spatiales naturellement hétérogènes : elles peuvent être écrites par différents auteurs, en utilisant un vocabulaire différent, à partir d'un point de vue différent. Elles peuvent par ailleurs couvrir des zones géographiques larges et diverses et contenir des niveaux de détail variés. Pour toutes ces raisons il est difficile d'intégrer dans les modèles de SIG l'information géographique provenant de texte. L'hypothèse du monde ouvert des technologies du Web sémantique induit que les graphes de connaissances sont une meilleure solution pour modéliser et stocker les connaissances géographiques extraites de textes hétérogènes, incomplets et imparfaits en langage naturel. Structurées en graphe de connaissances géospatial, les connaissances spatiales ambiguës peuvent être désambiguïsées et liées à des ressources géographiques de référence, ce qui les enrichit de références spatiales directes lorsque c'est possible et facilite considérablement leur accessibilité et réutilisation. L'objectif de cette thèse est de développer une approche opérationnelle pour la construction de graphes de connaissances à partir de texte et des données géographiques de référence. Cette approche doit permettre d'intégrer à la fois des références spatiales directes et indirectes. Nous appliquons nos recherches à un corpus de texte français, ce qui nous permet d'identifier et de valider empiriquement une méthodologie fonctionnelle pour la construction de graphes de connaissances géospatiales à partir de texte. Le corpus est constitué des ouvrages Instructions nautiques du Shom, qui décrivent l'environnement maritime côtier et donnent des instructions de navigation côtière. La contribution principale de cette thèse est la méthodologie ATONTE pour la construction semi-automatique de graphes de connaissances, géospatiaux ou non, à partir de texte, des connaissances d'experts et des données de référence. Nous présentons cette méthodologie en détail et nous démontrons la manière dont nous l'avons implémentée afin de construire un graphe de connaissances géospatial du contenu des Instructions nautiques. La première composante de la méthodologie ATONTE est une méthodologie pour le développement d'ontologies de domaine à partir de texte et d'experts. Nous l'appliquons à notre corpus, en intégrant les résultats d'entretiens réalisés auprès d'experts de notre corpus, afin de développer l'ontologie ATLANTIS : une ontologie noyau géospatiale du domaine des Instructions nautiques. La deuxième composante est une approche automatique pour l'extraction d'entités imbriquées et de relations binaires à partir de texte en utilisant des réseaux de neurones profonds. Les modèles sont entraînés sur un jeu de données annoté manuellement, spécifique au domaine. Nous implémentons cette approche afin d'extraire les entités et les relations spatiales de notre corpus, ce qui exige la création d'un jeu de données d'entraînement en français, annoté à la main. Nous donnons des résultats de référence pour ce jeu de données pour l'extraction d'entités spatiales imbriquées, l'extraction de relations spatiales binaires, et l'extraction combinée d'entités et de relations spatiales de bout en bout. La dernière composante utilise des outils disponibles afin de structurer les entités et relations spatiales extraites des Instructions nautiques selon l'ontologie ATLANTIS dans un premier temps, et de lier les entités à leurs entrées correspondantes dans la BD TOPO® dans un second temps. Le résultat est une base opérationnelle du graphe de connaissances géospatial des Instructions nautiques
Some spatial knowledge, current or historical, exists only in the form of text. Examples of such sources of unstructured spatial knowledge include travel guides, historical documents and social media posts. Textual sources contain naturally heterogeneous spatial knowledge: they can be written by different authors, using different vocabulary, from different points of view, they can cover large and diverse geographic areas, and they can contain varied levels of detail. These are some of the reasons why it is difficult to integrate geographic information from textual sources into GIS models, which require highly-structured complete data with direct spatial referencing. The open-world assumption of semantic Web technologies makes knowledge graphs a better solution for modelling and storing geographic information extracted from heterogeneous, incomplete and imperfect natural language text. Structured as a geospatial knowledge graph, what was once ambiguous spatial knowledge can be disambiguated and formally linked to reference geographic resources, thereby enriching it with direct spatial referencing where possible and significantly facilitating its accessibility and reuse.The objective of this thesis is to develop an operational approach for the creation of knowledge graphs from text and geographic reference data that is adapted to the special case of constructing geospatial knowledge graphs that include both direct and indirect spatial referencing. We apply our research to a French text corpus, which allows us to empirically identify and validate a functional methodology for creating geospatial knowledge graphs from text. The corpus is composed of the Instructions nautiques, a series of books published by the Shom that describe the maritime environment and give coastal navigation instructions.The main contribution of this thesis is the ATONTE Methodology for the semi-automatic construction and population of knowledge graphs, geospatial or not, from heterogeneous textual sources, expert knowledge and reference data. We present the ATONTE Methodology in detail and demonstrate how we implemented it to construct a geospatial knowledge graph of the content of the Instructions nautiques.The first of the three components that make up ATONTE is a novel methodology for the manual development of domain ontologies from text and the knowledge of domain experts. We apply this methodology to our corpus, integrating our findings from interviews carried out with expert users of the corpus, to develop the ATLANTIS Ontology: a geospatial seed ontology of the domain of the Instructions nautiques.The second component consists of a baseline approach for automatic nested entity and binary relation extraction from text using a deep neural network. It requires training two existing pretrained deep language models, one for the task of entity extraction and the other for relation extraction, on a domain-specific manually-annotated textual dataset. We implement the approach to extract the spatial entities and relations from our corpus, creating a French-language annotated training dataset in the process. We provide benchmark results for this dataset for three tasks: nested spatial entity extraction, binary spatial relation extraction, and end-to-end spatial entity and relation extraction. The third and final component is dedicated to automatically structuring the information extracted during the previous stage as a knowledge graph according to the ontology developed during the first stage, and disambiguating the entities via entity linking to a reference resource. We present a proof of concept of this stage, using off-the-shelf tools to first structure the spatial entities and relations extracted from the Instructions nautiques according to the ATLANTIS Ontology and then link the entities to their corresponding entries in the BD TOPO®. The result is an operational basis for the geospatial ATLANTIS Knowledge Graph of the Instructions nautiques

APA, Harvard, Vancouver, ISO, and other styles

15

Destandau, Marie. "Path-Based Interactive Visual Exploration of Knowledge Graphs." Electronic Thesis or Diss., université Paris-Saclay, 2020. http://www.theses.fr/2020UPASG063.

Full text

Abstract:

Les Graphes de connaissances représentent, connectent, et rendent interprétables par des algorithmes des connaissances issues de différents domaines. Ils reposent sur des énoncés simples que l’on peut chaîner pour former des énoncés de plus haut niveau. Produire des interfaces visuelles interactives pour explorer des collections dans ces données est un problème complexe, en grande partie non résolu. Dans cette thèse, je propose le concept de profils de chemins pour décrire les énoncés de haut niveau. Je l’utilise pour développer 3 outils open source : S-Paths permet de naviguer dans des collections à travers des vues synthétiques ; Path Outlines permet aux producteurs de données de parcourir les énoncés qui peuvent produits par leurs graphes ; et The Missing Path leur permet d’analyser l’incomplétude de leurs données. Je montre que le concept, en plus de supporter des interfaces visuelles interactives pour les graphes de connaissances, aide aussi à en améliorer la qualité
Knowledge Graphs facilitate the pooling and sharing of information from different domains. They rely on small units of information named triples that can be combined to form higher-level statements. Producing interactive visual interfaces to explore collections in Knowledge Graphs is a complex problem, mostly unresolved. In this thesis, I introduce the concept of path outlines to encode aggregate information relative to a chain of triples. I demonstrate 3 applications of the concept withthe design and implementation of 3 open source tools. S-Paths lets users browse meaningful overviews of collections; Path Outlines supports data producers in browsing the statements thatcan be produced from their data; and The Missing Path supports data producers in analysingincompleteness in their data. I show that the concept not only supports interactive visual interfaces for Knowledge Graphs but also helps better their quality

APA, Harvard, Vancouver, ISO, and other styles

16

Moniruzzaman, A. B. M. "Tensor modelling for fine-grained type entity inference in knowledge graphs." Thesis, Queensland University of Technology, 2021. https://eprints.qut.edu.au/208327/1/A%20B%20M_Moniruzzaman_Thesis.pdf.

Full text

Abstract:

Knowledge Graphs (KGs) are playing an increasingly important role in advancing the intelligence of the Web. Fine-grained type entity inferencing in a KG is very useful for enriching Semantic Web search results and allowing queries with a well-defined result set. This thesis developed two approaches based on tensor modelling for fine-grained type entity inference. In the first approach, it developed methods for utilising both embedded knowledge inside KGs and linked entity supplementary information outside KGs to improve inference accuracy. In the second approach, this thesis exploits type hierarchical path sampling technique to minimize the computational complexity of large-scale KG factorization.

APA, Harvard, Vancouver, ISO, and other styles

17

Zhang, Weijian. "Evolving graphs and similarity-based graphs with applications." Thesis, University of Manchester, 2018. https://www.research.manchester.ac.uk/portal/en/theses/evolving-graphs-and-similaritybased-graphs-with-applications(66a23d3d-1ad0-454b-9ba0-175b566af95d).html.

Full text

Abstract:

A graph is a mathematical structure for modelling the pairwise relations between objects. This thesis studies two types of graphs, namely, similarity-based graphs and evolving graphs. We look at ways to traverse an evolving graph. In particular, we examine the influence of temporal information on node centrality. In the process, we develop EvolvingGraphs.jl, a software package for analyzing time-dependent networks. We develop Etymo, a search system for discovering interesting research papers. Etymo utilizes both similarity-based graphs and evolving graphs to build a knowledge graph of research articles in order to help users to track the development of ideas. We construct content similarity-based graphs using the full text of research papers. And we extract key concepts from research papers and exploit the temporal information in research papers to construct a concepts evolving graph.

APA, Harvard, Vancouver, ISO, and other styles

18

Hadjidemetriou, Constantia. "Graphs : pupils understanding and teachers pedagogical content and knowledge." Thesis, University of Manchester, 2002. http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.631233.

Full text

Abstract:

This study explored 14 to 15 year old pupils' graphical understanding and their teachers' pedagogical content knowledge. A diagnostic instrument was developed from the research literature to suit the UK National Curriculum, and was administered to 425 pupils. The problems were deliberately posed in such a way as to encourage relevant errors and misconceptions to surface. The test was 'scaled' using Rasch methodology and the result was a hierarchy of responses, each level of which was described as a characteristic performance including key misconceptions. Results showed that pupils were able to solve tasks involving both reading and interpretation of graphs from an early level. The hierarchy was generally consistent with previous literature. The errors were validated apart from one which is believed to be a new version of the so called 'interval-point' confusion. The instrument was also further developed to function as a questionnaire for assessing teachers' Pedagogical Content Knowledge (PCK). Teachers' estimation of the difficulty of the items, their proposed learning sequences and their awareness of errors and misconceptions were examined. Furthermore, teachers' perceptions of what is difficult were correlated with the children's actual difficulty estimates. Results showed that these teachers' estimation of what is difficult seemed to be partly structured by the curriculum sequence. Some of the teachers overestimated the difficulty of some tasks involving global interpretation and underestimated the difficulty of those which entailed pointwise reading or algebraic manipulation. Also, their knowledge was highly sensitive to the method adopted to collect the data. The teachers' mis-estimation of (relative) difficulties could be explained by one of two reasons: sometimes teachers apparently misunderstood the actual question themselves, and so underestimated the difficulty of the item. At other times, teachers overestimated the difficulty because they did not realise that children could answer the question without a sophisticated understanding of some concepts. Pupils' and teachers' responses were confirmed and enriched through group interviews and semi-structured interviews respectively.

APA, Harvard, Vancouver, ISO, and other styles

19

Marshall, Oliver. "Search Engine Optimization and the connection with Knowledge Graphs." Thesis, Högskolan i Gävle, Företagsekonomi, 2021. http://urn.kb.se/resolve?urn=urn:nbn:se:hig:diva-35165.

Full text

Abstract:

Aim: The aim of this study is to analyze the usage of Search Engine Optimization and Knowledge Graphs and the connection between them to achieve profitable business visibility and reach. Methods: Following a qualitative method together with an inductive approach, ten marketing professionals were interviewed via an online questionnaire. To conduct this study both primary and secondary data was utilized. Scientific theory together with empirical findings were linked and discussed in the analysis chapter. Findings: This study establishes current Search Engine Optimization utilization by businesses regarding common techniques and methods. We demonstrate their effectiveness on the Google Knowledge Graph, Google My Business and resulting positive business impact for increased visibility and reach. Difficulties remain in accurate tracking procedures to analyze quantifiable results. Contribution of the thesis: This study contributes to the literature of both Search Engine Optimization and Knowledge Graphs by providing a new perspective on how these subjects have been utilized in modern marketing. In addition, this study provides an understanding of the benefits of SEO utilization on Knowledge Graphs. Suggestions for further research: We suggest more extensive investigation on the elements and utilization of Knowledge Graphs; how the structure can be affected; which techniques are most effective on a bigger scale and how effectively the benefits can be measured. Key Words: Search Engine, Search Engine Optimization, SEO, Knowledge Graphs, Google My Business, Google Search Engine, Online Marketing.

APA, Harvard, Vancouver, ISO, and other styles

20

Gottschalk, Simon [Verfasser]. "Creation, Enrichment and Application of Knowledge Graphs / Simon Gottschalk." Hannover : Gottfried Wilhelm Leibniz Universität, 2021. http://d-nb.info/1235138534/34.

Full text

APA, Harvard, Vancouver, ISO, and other styles

21

Dubey, Mohnish [Verfasser]. "Towards Complex Question Answering over Knowledge Graphs / Mohnish Dubey." Bonn : Universitäts- und Landesbibliothek Bonn, 2021. http://d-nb.info/1238687849/34.

Full text

APA, Harvard, Vancouver, ISO, and other styles

22

Zanella, Calzada Laura A. "Biomedical Event Extraction Based on Transformers and Knowledge Graphs." Electronic Thesis or Diss., Université de Lorraine, 2023. http://www.theses.fr/2023LORR0235.

Full text

Abstract:

L'extraction d'événements biomédicaux peut être divisée en trois sous-tâches principales : (1) la détection de déclencheurs d'événements biomédicaux, (2) l'identification d'arguments biomédicaux et (3) la construction d'événements. Dans cette étude, pour la première sous-tâche, nous analysons un ensemble de modèles de langage transformer couramment utilisés dans le domaine biomédical afin d'évaluer et de comparer leur capacité à détecter les déclencheurs d'événements. Nous affinons les modèles en utilisant sept corpus annotés manuellement pour évaluer leurs performances dans différents sous-domaines biomédicaux. SciBERT s'est révélé être le modèle le plus performant, présentant une légère amélioration par rapport aux modèles de référence. Pour la deuxième sous-tâche, nous construisons un graphe de connaissances (KG, en anglais) à partir des corpus biomédicaux et intégrons ses embeddings KG à SciBERT pour enrichir son information sémantique. Nous démontrons que l'ajout des embeddings KG au modèle améliore la performance de l'identification d'arguments d'environ 20 %, et d'environ 15 % par rapport à deux modèles de référence. Pour la troisième sous-tâche, nous utilisons le modèle génératif, ChatGPT, basé sur des invitations, pour construire l'ensemble final d'événements extraits. Nos résultats suggèrent que l'affinage d'un modèle transformateur pré-entraîné à partir de zéro avec des données biomédicales et générales permet de détecter les déclencheurs d'événements et d'identifier des arguments couvrant différents sous-domaines biomédicaux, améliorant ainsi sa généralisation. De plus, l'intégration des embeddings KG dans le modèle peut significativement améliorer la performance de l'identification d'arguments d'événements biomédicaux, surpassant les résultats des modèles de référence
Biomedical event extraction can be divided into three main subtasks; (1) biomedical event trigger detection, (2) biomedical argument identification and (3) event construction. In this work, for the first subtask, we analyze a set of transformer language models that are commonly used in the biomedical domain to evaluate and compare their capacity for event trigger detection. We fine-tune the models using seven manually annotated corpora to assess their performance in different biomedical subdomains. SciBERT emerged as the highest-performing model, presenting a slight improvement compared to baseline models. For the second subtask, we construct a knowledge graph (KG) from the biomedical corpora and integrate its KG embeddings to SciBERT to enrich its semantic information. We demonstrate that adding the KG embeddings to the model improves the argument identification performance by around 20 %, and by around 15 % compared to two baseline models. For the third subtask, we use the generative model, ChatGPT, based on prompts to construct the final set of extracted events. Our results suggest that fine-tuning a transformer model that is pre-trained from scratch with biomedical and general data allows to detect event triggers and identify arguments covering different biomedical subdomains, and therefore improving its generalization. Furthermore, the integration of KG embeddings into the model can significantly improve the performance of biomedical event argument identification, outperforming the results of baseline models

APA, Harvard, Vancouver, ISO, and other styles

23

Han, Kelvin. "Generating and answering questions across text and knowledge graphs." Electronic Thesis or Diss., Université de Lorraine, 2024. http://www.theses.fr/2024LORR0162.

Full text

Abstract:

La génération de questions (QG) est une tâche qui consiste à produire automatiquement une question à partir d'une source d'information en entrée contenant la réponse. Il s'agit d'une sous-tâche de la génération automatique de textes (NLG), elle est également liée à la tâche de questions-réponses (QA), qui est l'opposé de la QG. L'objectif de la QG est de générer une expression linguistique pour rechercher l'information, l'objectif du QA est d'identifier automatiquement la réponse à une question à partir d'une source d'information en entrée. Les deux tâches ont des applications dans des domaines tels que la recherche d'information, les dialogues et les conversations, et aussi dans l'éducation. Lorsque les tâches de QG et de QA sont tout deux utilisées pour évaluation de textes basées sur la QA, elles sont aussi utilisées pour la vérification des faits (notamment les sorties de la NLG qui peuvent être sur le résumé ou la génération de texte à partir des données). La plupart des recherches sur ces deux tâches se concentrent soit sur l'une soit sur l'autre, et généralement dans une seule et unique modalité. Dans le domaine de la QG, les approches antérieures reposaient sur des architectures nécessitant un prétraitement intensif. Les questions ainsi générées ne couvraient ni l'entièreté des informations en entrée, ni la diversité des nuances possibles. Dans le domaine des QA, bien que des approches aient été proposées pour répondre aux question à partir d'informations non structurées (par exemple, un document textuel brute), mais aussi structurées (par exemple, des graphes de connaissances (KG) ou des tableaux), ces méthodes ne sont pas transférables pour une autre modalité. Dans cette thèse, nous nous concentrons d'abord sur la QG, afin d'identifier les moyens de générer des questions à partir d'informations structurées et également non structurées, et de le faire de manière contrôlée pour augmenter la diversité et la couverture des questions générées. Ensuite, nous étudierons également la conduite de la QG et des QA par un modèle capable de générer des questions simples et complexes de manière contrôlée à partir d'une modalité, puis répondre sur une autre modalité. Enfin, nous examinerons la possibilité de faire la même tâche pour les langues avec peu de ressources autres que l'anglais, ce qui pourrait faciliter l'évaluation basée sur les QA pour ces langues
Question generation (QG) is the task of automatically producing a question given some information source containing the answer. It is a subtask within natural language generation (NLG) but is also closely associated with question answering (QA), which is a counterpoint to QG. While QG is concerned with generating the linguistic expression for seeking information, the QA task is concerned with meeting that need by automatically identifying the answer to a question given some information source. Both tasks have direct applicability in domains such as information retrieval, dialogue and conversation, and education. Recent research also indicates that QG and QA, when used jointly in QA-based evaluation, are helpful for factual verification (especially for NLG outputs such as summarisation and data-to-text generations). When used together to produce a discourse representation, they can also help reduce the propensity of large language models (LLMs) to produce text with hallucinations and factual inaccuracies. While QA has long been studied, and approaches have been proposed as early as the 1960s, QG only started to gain more research attention in recent years. Most research on the tasks is focused on addressing only one of them and doing so for a single modality. In QG, previous approaches typically rely on architectures that require heavy processing and do not generally consider the generation of questions across the entirety of the input information source nor the diversity of the ways a question can be phrased. In QA, although work has been done for answering questions given some unstructured input (e.g. a piece of text), and work has also been done for doing so given some structured input (e.g. knowledge graph (KG) or tables), these methods are typically not transferable for use on another input modality. In this thesis, we are focused on QG foremost, with the aim of identifying ways to generate questions across both structured and unstructured information, namely text and KG inputs,in a manner that is controllable for increasing the diversity, comprehensiveness, and coverageof these questions. We also study QG and QA in concert with a model that can controllably generate both simple and complex questions from one modality and also answer them on another modality, an ability that has relevance for improving QA-based evaluation. Finally, we examine doing so for lower-resourced languages other than English, with the view that being able to do so helps enable similar QA-based evaluation for these languages

APA, Harvard, Vancouver, ISO, and other styles

24

Nishioka, Chifumi [Verfasser]. "Profiling Users and Knowledge Graphs on the Web / Chifumi Nishioka." Kiel : Universitätsbibliothek Kiel, 2018. http://d-nb.info/115188071X/34.

Full text

APA, Harvard, Vancouver, ISO, and other styles

25

Mazibe, Ernest Nkosingiphile. "Teaching graphs of motion : translating pedagogical content knowledge into practice." Diss., University of Pretoria, 2017. http://hdl.handle.net/2263/62885.

Full text

Abstract:

This study investigated the comparison between captured and revealed Pedagogical Content Knowledge (PCK) about graphs of motion. The aim of the study was to explore PCK when captured in a written format and discussions (captured PCK) and compare it to the PCK that the same teachers revealed in practice (revealed PCK) when teaching the topic. Four Grade 10 Physical sciences teachers were purposively and conveniently selected as participants of the study. Their PCK was captured through Content Representations (CoRes) and interviews. The revealed PCK on the other hand was gathered through lesson observations. The Topic Specific Pedagogical Content Knowledge (TSPCK) model was used as the framework that guided the analysis of the two manifestations of PCK. The focus was on teachers’ competences in the TSPCK components namely; learners’ prior knowledge including misconceptions, curricular saliency, what is difficult to teach, representations including analogies, and conceptual teaching strategies. The results of this study indicated that teachers’ competences in the TSPCK components varied. This was evident in both the captured and the revealed PCK. Thus it suggested that a teacher’s level of competence in one component is not necessarily an indication of his or her competence in the other components that define PCK, and subsequently in his/her overall captured or revealed PCK. Furthermore, the study suggested that the level of competence in a component in the captured PCK is not necessarily an indication of the level of competence within that component that the teacher would reveal during lesson presentation. The level may be the same, slightly different (higher or lower) or even be drastically different in the lesson than suggested by the captured PCK. A concluding remark was then made that teachers’ captured PCK is not necessarily a true reflection of the PCK they reveal during lesson presentation and that different instruments must be used to reflect on and assess teachers’ PCK in a topic.
Dissertation (MEd)--University of Pretoria, 2017.
Science, Mathematics and Technology Education
MEd
Unrestricted

APA, Harvard, Vancouver, ISO, and other styles

26

González, Alejandro. "A Swedish Natural Language Processing Pipeline For Building Knowledge Graphs." Thesis, KTH, Skolan för elektroteknik och datavetenskap (EECS), 2018. http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-254363.

Full text

Abstract:

The concept of knowledge is proper only to the human being thanks to the faculty of understanding. The immaterial concepts, independent of the material causes of the experience constitute an evident proof of the existence of the rational soul that makes the human being a spiritual being "in a way independent of the material. Nowadays research efforts in the field of Artificial Intelligence are trying to mimic this human capacity using computers by means of tteachingthem how to read and understand human language using Machine Learning techniques related to the processing of human language. However, there are still a significant number of challenges such as how to represent this knowledge so can be used by a machine to infer conclusions or provide answers. This thesis presents a Natural Language Processing pipeline that is capable of building a knowledge representation of the information contained in Swedish human-generated text. The result is a system that, given Swedish text in its raw format, builds a representation in the form of a Knowledge Graph of the knowledge or information contained in that text.
Vetskapen om kunskap är den del av det som definierar den nutida människan (som vet, att hon vet). De immateriella begreppen oberoende av materiella attribut är en del av beviset på att människan en själslig varelse som till viss del är oberoende av materialet. För närvarande försöker forskningsinsatser inom artificiell intelligens efterlikna det mänskliga betandet med hjälp av datorer genom att "lära" dem hur man läser och förstår mänskligt språk genom att använda maskininlärningstekniker relaterade till behandling av mänskligt språk. Det finns emellertid fortfarande ett betydande antal utmaningar, till exempel hur man representerar denna kunskap så att den kan användas av en maskin för att dra slutsatser eller ge svar utifrån detta. Denna avhandling presenterar en studie i användningen av ”Natural Language Processing” i en pipeline som kan generera en kunskapsrepresentation av informationen utifrån det svenska språket som bas. Resultatet är ett system som, med svensk text i råformat, bygger en representation i form av en kunskapsgraf av kunskapen eller informationen i den texten.

APA, Harvard, Vancouver, ISO, and other styles

27

Tailhardat, Lionel. "Anomaly detection using knowledge graphs and synergistic reasoning : application to network management and cyber security." Electronic Thesis or Diss., Sorbonne université, 2024. http://www.theses.fr/2024SORUS293.

Full text

Abstract:

La gestion des incidents sur les réseaux informatiques et télécoms, qu'il s'agisse de problèmes d'infrastructure ou de cybersécurité, nécessite la capacité de traiter et d'interpréter simultanément et rapidement un grand nombre de sources d'information techniques hétérogènes en matière de format et de sémantique. Dans cette thèse, nous étudions l'intérêt de structurer ces données dans un graphe de connaissances, et étudions en quoi cette structure permet de maîtriser la complexité des réseaux, notamment pour des applications en détection d'anomalies sur des réseaux dynamiques et de grande taille. À travers une ontologie (un modèle des concepts et relations pour décrire un domaine d'application), les graphes de connaissances permettent en effet de donner un sens commun à des informations différentes en apparence. Nous introduisons pour commencer une nouvelle ontologie permettant de décrire les infrastructures réseaux, les incidents, et les opérations d'exploitation. Nous décrivons de même une architecture pour transformer les données des réseaux en un graphe de connaissance organisé selon cette ontologie, en utilisant les techniques du Web Sémantique pour favoriser l'interopérabilité. Le graphe de connaissance résultant permet d'analyser le comportement des réseaux de façon homogène. Nous définissons ensuite trois familles de techniques algorithmiques pour utiliser les données du graphe, et montrons comment ces techniques peuvent être utilisées pour détecter des comportements anormaux des systèmes et aider les équipes d'exploitation dans le diagnostic d'incidents. Enfin, nous présentons une architecture logicielle pour simplifier les interactions des exploitants avec le graphe de connaissance et les algorithmes d'aide au diagnostique par l'intermédiaire d'une interface graphique spécialisée. Chaque proposition a été testé de manière indépendante par des expérimentations et démonstrations, ainsi que par un panel d'utilisateurs experts depuis l'interface graphique spécialisée dans le cadre d'une solution intégrée
Incident management on telecom and computer networks, whether it is related to infrastructure or cybersecurity issues, requires the ability to simultaneously and quickly correlate and interpret a large number of heterogeneous technical information sources. In this thesis, we study the benefits of structuring this data into a knowledge graph, and examine how this structure helps to manage the complexity of networks, particularly for applications in anomaly detection on dynamic and large-scale networks. Through an ontology (a model of concepts and relationships to describe an application domain), knowledge graphs allow different-looking information to be given a common meaning. We first introduce a new ontology for describing network infrastructures, incidents, and operations. We also describe an architecture for transforming network data into a knowledge graph organized according to this ontology, using Semantic Web technologies to foster interoperability. The resulting knowledge graph allows for standardized analysis of network behavior. We then define three families of algorithmic techniques for using the graph data, and show how these techniques can be used to detect abnormal system behavior and assist technical support teams in incident diagnosis. Finally, we present a software architecture to facilitate the interactions of support teams with the knowledge graph and diagnostic algorithms through a specialized graphical user interface. Each proposal has been independently tested through experiments and demonstrations, as well as by a panel of expert users using the specialized graphical interface within an integrated solution

APA, Harvard, Vancouver, ISO, and other styles

28

Hur, Muhammad Ali. "A framework to support autonomous construction of knowledge graphs from unstructured text." Thesis, Edith Cowan University, Research Online, Perth, Western Australia, 2024. https://ro.ecu.edu.au/theses/2801.

Full text

Abstract:

This thesis delves into the automated construction of Knowledge Graphs (KGs) from unstructured text data, aiming to overcome the challenges inherent in extracting and representing contextual information with the necessary semantic depth and breadth. While various approaches exist for extracting semantic relations such as temporal, causal, and rhetorical from unstructured text, they often focus on one type of relation at the expense of others, resulting in a fragmented contextual representation. Consequently, these approaches fail to produce KGs with rich semantic information, hindering the development of advanced AI applications like recommendation engines and semantic search. Moreover, existing approaches prioritize data integration over quality improvement, neglecting critical aspects such as accuracy, consistency, and completeness, which can lead to errors and inconsistencies in the resulting KGs. To address these issues, this research proposes a comprehensive framework that integrates sophisticated semantic models and linguistic analysis techniques to enhance the depth and precision of semantic representations within KGs. By leveraging a graph-based approach, the proposed framework captures diverse semantic relationships and contextual cues present within unstructured text, providing a structured foundation for the seamless integration and interpretation of textual information. The contributions of this research include the development of semantic enrichment techniques, unified context representation frameworks, and advanced semantic analysis models. These contributions enable the extraction and representation of semantic relations at various linguistic levels, including morphological, syntactic, and semantic aspects. Furthermore, the research explores the practical implications of the proposed framework, demonstrating its utility in various domains such as natural language processing, information retrieval, and knowledge management. The framework undergoes validation utilizing a gold-standard MEANTIME dataset to ensure its efficacy in domain-agnostic text representation. This validation assesses the accuracy of semantic elements including events, entities, event participants, temporal relations, and coreference links. Additionally, for the validation of domain-specific KGs, a carefully crafted domain-specific stock market ontology and a set of competency questions serve as benchmarks against which the domain-specific KG is rigorously evaluated and validated. Overall, this thesis contributes to advancing the state-of-the-art in automated knowledge extraction from unstructured ii text data, paving the way for more informed decision-making and sophisticated information processing systems.

APA, Harvard, Vancouver, ISO, and other styles

29

Hanika, Tom [Verfasser]. "Discovering Knowledge in Bipartite Graphs with Formal Concept Analysis / Tom Hanika." Kassel : Universitätsbibliothek Kassel, 2019. http://d-nb.info/1180660811/34.

Full text

APA, Harvard, Vancouver, ISO, and other styles

30

Santos, Henrique Oliveira. "An indicator-based approach for variable alignment based on knowledge graphs." Universidade de Fortaleza, 2018. http://dspace.unifor.br/handle/tede/107852.

Full text

Abstract:

Made available in DSpace on 2019-03-30T00:02:07Z (GMT). No. of bitstreams: 0 Previous issue date: 2018-08-30
Scientific data is being generated and acquired in high volumes in support of studies in many domain areas. In current scenarios, data files containing values of variables (scientific measurements and/or study objects), are ultimately leveraged by data scientists in a series of data preparation tasks that aim to identify relationships between variables in a way that they can be reorganized in an aligned manner, e.g., rewritten as a single line in a tabular file following an alignment criterion. This criterion plays the role of a relationship between a number of distinct variables that is not trivial or easy to elicit looking directly into data files. To address this challenge, we propose a workflow for scientific data characterization and variable alignment based on user-defined indicators. The workflow is able to semantically characterize tabular scientific data files using scientific and domain knowledge in knowledge graphs, allowing data to be queried and retrieved by an ontology-driven faceted-search. A representation of indicators that mimics data users' comparisons and visualizations needs is then leveraged by tasks that are able to produce aligned datasets that can be used directly in routine data tools like R or business intelligence (BI) software for easy graphical plotting. We demonstrate the execution of the workflow in the context of two use cases using data files from the city of Fortaleza, Brazil, where an implementation of this work was used by identified stakeholders. During rounds of evaluation, our approach was verified to ease the process of extracting insights and visualization from scientific data files. To conclude, we discuss the outcomes of this work and their impact on the existing literature, showing ongoing work and potential research directions. Keywords Knowledge graphs; scientific data; data analysis; variable alignment; indicators
Dados científicos são gerados e adquiridos em grandes volumes em apoio a estudos em diversas áreas do conhecimento. Processos de preparação de dados comumente usados fazem uso desses arquivos de dados científicos com a finalidade de identificar relacionamentos implícitos entre variáveis de tal forma que eles possam ser reorganizados de forma alinhada, i.e., reescritos como uma única linha em um arquivo tabular seguindo um critério de alinhamento. Esse critério tem o papel de um relacionamento entre variáveis diversas que não é trivial ou fácil de se extrair verificando diretamente nos arquivos de dados. Para enfrentar esse desafio, propomos um fluxo de trabalho para a caracterização de dados científicos e alinhamento de variáveis baseado na definição de indicadores por usuários dos dados. O fluxo de trabalho tem a capacidade de caracterizar semanticamente arquivos tabulares contendo dados científicos utilizando conhecimento científico e de domínio presente em grafos de conhecimento, permitindo que os dados sejam consultados e recuperados através de uma busca facetada guiada por ontologias. Uma representação de indicadores que reproduz as necessidades de comparações e visualizações de variáveis de usuários dos dados é utilizada para se produzir conjunto de dados alinhados que podem ser utilizados diretamente em ferramentas de dados existentes, como R ou soluções de business intelligence (BI) para plotagem gráfica de modo fácil. Nós demonstramos a execução do fluxo de trabalho no contexto de dois casos de uso utilizando arquivos de dados da cidade de Fortaleza, Brasil, onde uma implementação desse trabalho foi utilizada por partes interessadas. Durante rodadas de avaliação, nossa proposta foi verificada como facilitadora do processo de extração de visões gerais, percepções e visualizações a partir de arquivos de dados científicos. Em conclusão, nós discutimos os resultados desse trabalho e seu impacto na literatura existente, mostrando trabalhos em andamento e potenciais direções de pesquisa. Palavras-chave Grafos de conhecimento; dados científios; análise de dados; alinhamento de variáveis; indicadores

APA, Harvard, Vancouver, ISO, and other styles

31

Arrascue, Ayala Victor Anthony [Verfasser], and Georg [Akademischer Betreuer] Lausen. "Towards an effective consumption of large-scale knowledge graphs for recommendations." Freiburg : Universität, 2020. http://d-nb.info/1223366189/34.

Full text

APA, Harvard, Vancouver, ISO, and other styles

32

Nguyen, Vinh Thi Kim. "Semantic Web Foundations for Representing, Reasoning, and Traversing Contextualized Knowledge Graphs." Wright State University / OhioLINK, 2017. http://rave.ohiolink.edu/etdc/view?acc_num=wright1516147861789615.

Full text

APA, Harvard, Vancouver, ISO, and other styles

33

Wiradee, Imrattanatrai. "Supporting Entity-oriented Search with Fine-grained Information in Knowledge Graphs." Kyoto University, 2020. http://hdl.handle.net/2433/259074.

Full text

APA, Harvard, Vancouver, ISO, and other styles

34

Janson, Axel, and Rietz Marc Du. "Knowledge-Based Expansions for Strategy Synthesis in Discrete Games on Graphs." Thesis, KTH, Skolan för elektroteknik och datavetenskap (EECS), 2020. http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-297693.

Full text

Abstract:

When analyzing situations involving intelligent agents with objectives, it can be helpful to use discrete games as models. Within such game models the synthesis of winning strategies is of interest, and many algorithmic methods have been developed for this purpose. This project focused on a less tractable game type, involving a coalition of players without the ability to communicate. For this type of game we propose two methods for exploring knowledge-based strategies. One is an extension of the previously developed Multiplayer Knowledge- Based Subset Construction with the additional assumption of action observability within the coalition. The other is a novel method called Epistemic Expansion, which assumes that the coalition coordinates before playing the game. We demonstrate how these methods can be used to help find winning strategies in example games with relevant properties.
Diskreta spel kan utgöra lämpliga modeller för många situationer där intelligenta spelare är inblandade. I dessa modellspel är det av intresse att hitta vinnande strategier och många algoritmer har utvecklats i detta syfte. I detta projekt undersöker vi spel i vilka en koalition av spelare med ett gemensamt mål ska samarbeta utan kommunikation. För att underlätta syntesen av vinnande strategier är det lämpligt att följa hur spelarnas kunskapsläge utvecklas under spelets gång. Vi föreslår två kunskapsbaserade konstruktioner för att modellera två skilda antaganden. Det första är att spelarna kan observera varandras handlingar, och det andra är att koalitionen har möjlighet att skapa en koordinerad strategi innan spelet börjar. Vi visar hur dessa konstruktioner kan användas för att syntetisera vinnande strategier i utvalda exempelspel.
Kandidatexjobb i elektroteknik 2020, KTH, Stockholm

APA, Harvard, Vancouver, ISO, and other styles

35

Zafartavanaelmi, Hamid [Verfasser]. "Semantic Question Answering Over Knowledge Graphs: Pitfalls and Pearls / Hamid Zafartavanaelmi." Bonn : Universitäts- und Landesbibliothek Bonn, 2021. http://d-nb.info/1238687393/34.

Full text

APA, Harvard, Vancouver, ISO, and other styles

36

Campos, Moussallem Diego [Verfasser]. "Knowledge graphs for multilingual language translation and generation / Diego Campos Moussallem." Paderborn : Universitätsbibliothek, 2020. http://d-nb.info/1213802822/34.

Full text

APA, Harvard, Vancouver, ISO, and other styles

37

Cabrera, Christian Bernabe <1992&gt. "Querying and Clustering on Knowledge Graphs: A dominant-set based approach." Master's Degree Thesis, Università Ca' Foscari Venezia, 2021. http://hdl.handle.net/10579/18811.

Full text

Abstract:

The thesis is part of the Horizon 2020 MEMEX project. MEMEX is a social project that promotes cultural heritage and social inclusion focusing on the minor communities of our society. This would be accomplished by the research and development of new technologies. The main treated topics are knowledge graphs (KG), localization based on computer vision and augmented reality. In this thesis we will focus on KG. This kind of graphs are a collection of interlinked entities and different types of edges. Each edge, with its description, represents a formal semantic. Entities, throughout descriptions and their attributes, contribute to each other creating a network which can be easily interpreted. KG in literature are not a new concept, however only in the recent years have acquired relevance in the development of new technologies. Our work will focus on their analysis, specifically in the study of clustering techniques that take into account both entities attributes and relationships semantics.

APA, Harvard, Vancouver, ISO, and other styles

38

BIANCHI, FEDERICO. "Corpus-based Comparison of Distributional Models of Language and Knowledge Graphs." Doctoral thesis, Università degli Studi di Milano-Bicocca, 2020. http://hdl.handle.net/10281/263553.

Full text

Abstract:

L'intelligenza artificiale cerca di spiegare come gli agenti intelligenti si comportano. Il linguaggio è uno dei media di comunicazioni più importanti e studiare delle teorie che permettano di definire il significato di espressioni naturali è molto importante. I linguisti hanno usato con successo linguaggi artificiali basati su logiche, ma una theory che ha avuto un impatto significativo in intelligenza artificiale è la semantica distribuzionale. La semantica distribuzionale afferma che il significato di espressioni in linguaggio naturale può essere derivato dal contesto in cui tali espressioni compaiono. Questa teoria è stata implementata da algoritmi che permettono di generare rappresentazioni vettoriali delle espressioni del linguaggio natural in modo che espressioni simili vengano rappresentate con vettori simili. Negli ultimi anni, gli scienziati cognitivi hanno sottolineato che queste rappresentazioni sono correlate con l'associative learning e che sono anche in grado di catturare bias e stereotype del testo. Diventa quindi importante trovare metodologie per comparare rappresentazioni che arrivano da sorgenti diverse. Ad esempio, usare questi algoritmi su testi di periodi differenti genera rappresentazioni differenti: visto che il linguaggio muta nel tempo, trovare delle metododoloie per comparare come le parole si sono mosse è un task imporante per l'intelligenza artificiale (e.g., la parola "amazon" ha cambiato il suo significato principale negli ultimi anni) In questa tesi, introduciamo un modello comparative basato su testi che permette di comparare rappresentazioni di sorgenti diverse generate con la semantica distribuzionale. Proponiamo un modello che è efficiente ed efficace e mostriamo che possiamo anche gestire nomi di entità e non solo paorle, superando problemi legati all'ambiguità del linguaggio. Alla fine, mostriamo che è possibile combinare questi metodi con approcci logici e fare comparazioni utilizzando costrutti logici.
One of the main goals of artificial intelligence is understanding how intelligent agent acts. Language is one of the most important media of communication, and studying theories that can account for the meaning of natural language expressions is an important task. Language is one of the most important media of communication, and studying theories that can account for the meaning of natural language expressions is a crucial task in artificial intelligence. Distributional semantics states that the meaning of natural language expressions can be derived from the context in which the expressions appear. This theory has been implemented by algorithms that generate vector representations of natural language expressions that represent similar natural language expressions with similar vectors. In the last years, several cognitive scientists have shown that these representations are correlated with associative learning and they capture cognitive biases and stereotypes as they are encoded in text corpora. If language is encoding important aspects of cognition and our associative knowledge, and language usage change across the contexts, the comparison of language usage in different contexts may reveal important associative knowledge patterns. Thus, if we want to reveal these patterns, we need ways to compare distributional representations that are generated from different text corpora. For example, using these algorithms on textual documents from different periods will generate different representations: since language evolves during time, finding a way to compare words that have shifted over time is a valuable task for artificial intelligence (e.g., the word "Amazon" has changed its prevalent meaning during the last years). In this thesis, we introduce a corpus-based comparative model that allows us to compare representations of different sources generated under the distributional semantic theory. We propose a model that is both effective and efficient, and we show that it can also deal with entity names and not just words, overcoming some problems that follow from the ambiguity of natural language. Eventually, we combine these methods with logical approaches. We show that we can do logical reasoning on these representations and make comparisons based on logical constructs.

APA, Harvard, Vancouver, ISO, and other styles

39

Heaton, John Edward. "Goal driven theorem proving using conceptual graphs and Peirce logic." Thesis, Loughborough University, 1994. https://dspace.lboro.ac.uk/2134/7706.

Full text

Abstract:

The thesis describes a rational reconstruction of Sowa's theory of Conceptual Graphs. The reconstruction produces a theory with a firmer logical foundation than was previously the case and which is suitable for computation whilst retaining the expressiveness of the original theory. Also, several areas of incompleteness are addressed. These mainly concern the scope of operations on conceptual graphs of different types but include extensions for logics of higher orders than first order. An important innovation is the placing of negation onto a sound representational basis. A comparison of theorem proving techniques is made from which the principles of theorem proving in Peirce logic are identified. As a result, a set of derived inference rules, suitable for a goal driven approach to theorem proving, is developed from Peirce's beta rules. These derived rules, the first of their kind for Peirce logic and conceptual graphs, allow the development of a novel theorem proving approach which has some similarities to a combined semantic tableau and resolution methodology. With this methodology it is shown that a logically complete yet tractable system is possible. An important result is the identification of domain independent heuristics which follow directly from the methodology. In addition to the theorem prover, an efficient system for the detection of selectional constraint violations is developed. The proof techniques are used to build a working knowledge base system in Prolog which can accept arbitrary statements represented by conceptual graphs and test their semantic and logical consistency against a dynamic knowledge base. The same proof techniques are used to find solutions to arbitrary queries. Since the system is logically complete it can maintain the integrity of its knowledge base and answer queries in a fully automated manner. Thus the system is completely declarative and does not require any programming whatever by a user with the result that all interaction with a user is conversational. Finally, the system is compared with other theorem proving systems which are based upon Conceptual Graphs and conclusions about the effectiveness of the methodology are drawn.

APA, Harvard, Vancouver, ISO, and other styles

40

Ceroni, Samuele. "Time-evolving knowledge graphs based on Poirot: dynamic representation of patients' voices." Bachelor's thesis, Alma Mater Studiorum - Università di Bologna, 2021. http://amslaurea.unibo.it/23095/.

Full text

Abstract:

Nowadays people are spending more and more time online: this is a permanent change that leads to a huge amount of diversified data like never before which needs to be managed to extrapolate knowledge from it. This also involves social media which produces free textual information very difficult to process, but occasionally very useful. For instance, in the field of rare diseases, our specific testing context could lead to the possibility to organize the voice of patients and of caregivers, difficult to gather otherwise. People who are affected by a rare disease often strive to find enough information about it. Indeed, not much material is available online and the number of doctors qualified for those specific diseases is quite limited. Social networks become then the best place to exchange ideas and opinions. The main difficulty in finding useful information on social networks though is that text gets lost quickly and it's not straightforward to give a semantic structure to it and dynamically evolve this representation over time. In literature, there are some techniques that manage to transform unstructured data into useful information, extracting them using artificial intelligence. These techniques are often well expressive and are able to precisely convert data into knowledge, but they are not directly connected to text sources nor to a system that stores and allows to update the extrapolated information. Consequently, they are not well automated in incrementally keeping information up-to-date as new text is provided, resulting in the need for a mechanical process to do it. The contribution proposed in this thesis focuses on how to use these technologies to maintain information in order over time, enhancing their usability and freshness. It consists of a system that connects the text source providers to the built knowledge graph, which contains the knowledge acquired and updated.

APA, Harvard, Vancouver, ISO, and other styles

41

Gao, Xiaoxu. "Exploring declarative rule-based probabilistic frameworks for link prediction in Knowledge Graphs." Thesis, KTH, Skolan för informations- och kommunikationsteknik (ICT), 2017. http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-210650.

Full text

Abstract:

En kunskapsgraf lagrar information från webben i form av relationer mellan olika entiteter. En kunskapsgrafs kvalité bestäms av hur komplett den är och dess noggrannhet. Dessvärre har många nuvarande kunskapsgrafer brister i form av saknad fakta och inkorrekt information. Nuvarande lösningar av länkförutsägelser mellan entiteter har problem med skalbarhet och hög arbetskostnad. Denna uppsats föreslår ett deklarativt regelbaserat probabilistiskt ramverk för att utföra länkförutsägelse. Systemet involverar en regelutvinnande modell till ett “hinge-loss Markov random fields” för att föreslå länkar. Vidare utvecklades tre strategier för regeloptimering för att förbättra reglernas kvalité. Jämfört med tidigare lösningar så bidrar detta arbete till att drastiskt reducera arbetskostnader och en mer spårbar modell. Varje metod har utvärderas med precision och F-värde på NELL och Freebase15k. Det visar sig att strategin för regeloptimering presterade bäst. MAP-uppskattningen för den bästa modellen på NELL är 0.754, vilket är bättre än en nuvarande spjutspetsteknologi graphical model(0.306). F-värdet för den bästa modellen på Freebase15k är 0.709.
The knowledge graph stores factual information from the web in form of relationships between entities. The quality of a knowledge graph is determined by its completeness and accuracy. However, most current knowledge graphs often miss facts or have incorrect information. Current link prediction solutions have problems of scalability and high labor costs. This thesis proposed a declarative rule-based probabilistic framework to perform link prediction. The system incorporates a rule-mining model into a hingeloss Markov random fields to infer links. Moreover, three rule optimization strategies were developed to improve the quality of rules. Compared with previous solutions, this work dramatically reduces manual costs and provides a more tractable model. Each proposed method has been evaluated with Average Precision or F-score on NELL and Freebase15k. It turns out that the rule optimization strategy performs the best. The MAP of the best model on NELL is 0.754, better than a state-of-the-art graphical model (0.306). The F-score of the best model on Freebase15k is 0.709.

APA, Harvard, Vancouver, ISO, and other styles

42

Sherwell, Brian W. "EXPLANATIONS IN CONTEXTUAL GRAPHS: A SOLUTION TO ACCOUNTABILITY IN KNOWLEDGE BASED SYSTEMS." Master's thesis, University of Central Florida, 2005. http://digital.library.ucf.edu/cdm/ref/collection/ETD/id/3050.

Full text

Abstract:

In order for intelligent systems to be a viable and utilized tool, a user must be able to understand how the system comes to a decision. Without understanding how the system arrived at an answer, a user will be less likely to trust its decision. One way to increase a user's understanding of how the system functions is by employing explanations to account for the output produced. There have been attempts to explain intelligent systems over the past three decades. However, each attempt has had shortcomings that separated the logic used to produce the output and that used to produce the explanation. By using the representational paradigm of Contextual Graphs, it is proposed that explanations can be produced to overcome these shortcomings. Two different temporal forms of explanations are proposed, a pre-explanation and a post-explanation. The pre-explanation is intended to help the user understand the decision making process. The post-explanation is intended to help the user understand how the system arrived at a final decision. Both explanations are intended to help the user gain a greater understanding of the logic used to compute the system's output, and thereby enhance the system's credibility and utility. A prototype system is constructed to be used as a decision support tool in a National Science Foundation research program. The researcher has spent the last year at the NSF collecting the knowledge implemented in the prototype system.
M.S.Cp.E.
Department of Electrical and Computer Engineering
Engineering and Computer Science
Computer Engineering

APA, Harvard, Vancouver, ISO, and other styles

43

McNaughton, Ross. "Inference graphs : a structural model and measures for evaluating knowledge-based systems." Thesis, London South Bank University, 1995. http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.260994.

Full text

APA, Harvard, Vancouver, ISO, and other styles

44

PORRINI, RICCARDO. "Construction and Maintenance of Domain Specific Knowledge Graphs for Web Data Integration." Doctoral thesis, Università degli Studi di Milano-Bicocca, 2016. http://hdl.handle.net/10281/126789.

Full text

Abstract:

A Knowledge Graph (KG) is a semantically organized, machine readable collection of types, entities, and relations holding between them. A KG helps in mitigating semantic heterogeneity in scenarios that require the integration of data from independent sources into a so called dataspace, realized through the establishment of mappings between the sources and the KG. Applications built on top of a dataspace provide advanced data access features to end-users based on the representation provided by the KG, obtained through the enrichment of the KG with domain specific facets. A facet is a specialized type of relation that models a salient characteristic of entities of particular domains (e.g., the vintage of wines) from an end-user perspective. In order to enrich a KG with a salient and meaningful representation of data, domain experts in charge of maintaining the dataspace must be in possess of extensive knowledge about disparate domains (e.g., from wines to football players). From an end-user perspective, the difficulties in the definition of domain specific facets for dataspaces significantly reduce the user-experience of data access features and thus the ability to fulfill the information needs of end-users. Remarkably, this problem has not been adequately studied in the literature, which mostly focuses on the enrichment of the KG with a generalist, coverage oriented, and not domain specific representation of data occurring in the dataspace. Motivated by this challenge, this dissertation introduces automatic techniques to support domain experts in the enrichment of a KG with facets that provide a domain specific representation of data. Since facets are a specialized type of relations, the techniques proposed in this dissertation aim at extracting salient domain specific relations. The fundamental components of a dataspace, namely the KG and the mappings between sources and KG elements, are leveraged to elicitate such domain specific representation from specialized data sources of the dataspace, and to support domain experts with valuable information for the supervision of the process. Facets are extracted by leveraging already established mappings between specialized sources and the KG. After extraction, a domain specific interpretation of facets is provided by re-using relations already defined in the KG, to ensure tight integration of data. This dissertation introduces also a framework to profile the status of the KG, to support the supervision of domain experts in the above tasks. Altogether, the contributions presented in this dissertation provide a set of automatic techniques to support domain experts in the evolution of the KG of a dataspace towards a domain specific, end-user oriented representation. Such techniques analyze and exploit the fundamental components of a dataspace (KG, mappings, and source data) with an effectiveness not achievable with state-of-the-art approaches, as shown by extensive evaluations conducted in both synthetic and real world scenarios.

APA, Harvard, Vancouver, ISO, and other styles

45

Paris, Pierre-Henri. "Identity in RDF knowledge graphs : propagation of properties between contextually identical entities." Electronic Thesis or Diss., Sorbonne université, 2020. http://www.theses.fr/2020SORUS132.

Full text

Abstract:

En raison du grand nombre de graphes de connaissances et, surtout, de leurs interconnexions encore plus nombreuses à l'aide de la propriété owl:sameas, il est devenu de plus en plus évident que cette propriété est souvent mal utilisée. En effet, les entités liées par la propriété owl:sameas doivent être identiques dans tous les contextes possibles et imaginables. Dans les faits, ceci n'est pas toujours le cas et induit une détérioration de la qualité des données. L'identité doit être considérée comme étant dépendante d'un contexte. Nous avons donc proposé une étude à large échelle sur la présence de la sémantique dans les graphes de connaissances, puisque certaines caractéristiques sémantiques permettent justement de déduire des liens d'identités. Cette étude nous a amenés naturellement à construire une ontologie permettant de donner la teneur en sémantique d'un graphe de connaissances. Nous avons aussi proposé une approche de liage de données fondée à la fois sur la logique permise par les définitions sémantiques, et à la fois sur la prédominance de certaines propriétés pour caractériser la relation d'identité entre deux entités. Nous nous sommes aussi intéressés à la complétude et avons proposé une approche permettant de générer un schéma conceptuel afin de mesurer la complétude d'une entité. Pour finir, à l'aide des travaux précédents, nous avons proposé une approche fondée sur les plongements de phrases permettant de calculer les propriétés pouvant être propagées dans un contexte précis. Ceci permet l'expansion de requêtes SPARQL et, in fine, d'augmenter la complétude des résultats de la requête
Due to a large number of knowledge graphs and, more importantly, their even more numerous interconnections using the owl:sameas property, it has become increasingly evident that this property is often misused. Indeed, the entities linked by the owl:sameas property must be identical in all possible and imaginable contexts. This is not always the case and leads to a deterioration of data quality. Identity must be considered as context-dependent. We have, therefore, proposed a large-scale study on the presence of semantics in knowledge graphs since specific semantic characteristics allow us to deduce identity links. This study naturally led us to build an ontology allowing us to describe the semantic content of a knowledge graph. We also proposed a interlinking approach based both on the logic allowed by semantic definitions, and on the predominance of certain properties to characterize the identity relationship between two entities. We looked at completeness and proposed an approach to generate a conceptual schema to measure the completeness of an entity. Finally, using our previous work, we proposed an approach based on sentence embedding to compute the properties that can be propagated in a specific context. Hence, the propagation framework allows the expansion of SPARQL queries and, ultimately, to increase the completeness of query results

APA, Harvard, Vancouver, ISO, and other styles

46

Chittella, Rama Someswar. "Leveraging Schema Information For Improved Knowledge Graph Navigation." Wright State University / OhioLINK, 2019. http://rave.ohiolink.edu/etdc/view?acc_num=wright1564755327091243.

Full text

APA, Harvard, Vancouver, ISO, and other styles

47

Adjei, Seth Akonor. "Refining Prerequisite Skill Structure Graphs Using Randomized Controlled Trials." Digital WPI, 2018. https://digitalcommons.wpi.edu/etd-dissertations/177.

Full text

Abstract:

Prerequisite skill structure graphs represent the relationships between knowledge components. Prerequisite structure graphs also propose the order in which students in a given curriculum need to be taught specific knowledge components in order to assist them build on previous knowledge and improve achievement in those subject domains. The importance of accurate prerequisite skill structure graphs can therefore not be overemphasized. In view of this, many approaches have been employed by domain experts to design and implement these prerequisite structures. A number of data mining techniques have also been proposed to infer these knowledge structures from learner performance data. These methods have achieved varied degrees of success. Moreover, to the best of our knowledge, none of the methods have employed extensive randomized controlled trials to learn about prerequisite skill relationships among skills. In this dissertation, we motivate the need for using randomized controlled trials to refine prerequisite skill structure graphs. Additionally, we present PLACEments, an adaptive testing system that uses a prerequisite skill structure graph to identify gaps in studentsâ€™ knowledge. Students with identified gaps are assisted with more practice assignments to ensure that the gaps are closed. PLACEments additionally allows for randomized controlled experiments to be performed on the underlying prerequisite skill structure graph for the purpose of refining the structure. We present some of the different experiment categories which are possible in PLACEments and report the results of one of these experiment categories. The ultimate goal is to inform domain experts and curriculum designers as they create policies that govern the sequencing and pacing of contents in learning domains whose content lend themselves to sequencing. By extension students and teachers who apply these policies benefit from the findings of these experiments.

APA, Harvard, Vancouver, ISO, and other styles

48

Barrière, Caroline. "From a children's first dictionary to a lexical knowledge base of conceptual graphs." Thesis, National Library of Canada = Bibliothèque nationale du Canada, 1997. http://www.collectionscanada.ca/obj/s4/f2/dsk3/ftp04/nq24293.pdf.

Full text

APA, Harvard, Vancouver, ISO, and other styles

49

Ristoski, Petar Verfasser], and Heiko [Akademischer Betreuer] [Paulheim. "Exploiting semantic web knowledge graphs in data mining / Petar Ristoski ; Betreuer: Heiko Paulheim." Mannheim : Universitätsbibliothek Mannheim, 2018. http://d-nb.info/1151446785/34.

Full text

APA, Harvard, Vancouver, ISO, and other styles

50

Ristoski, Petar [Verfasser], and Heiko [Akademischer Betreuer] Paulheim. "Exploiting semantic web knowledge graphs in data mining / Petar Ristoski ; Betreuer: Heiko Paulheim." Mannheim : Universitätsbibliothek Mannheim, 2018. http://nbn-resolving.de/urn:nbn:de:bsz:180-madoc-437307.

Full text

APA, Harvard, Vancouver, ISO, and other styles

Dissertations / Theses on the topic 'Knowledge Graphs'

Create a spot-on reference in APA, MLA, Chicago, Harvard, and other styles