Academic literature on the topic 'Triplestore databáze'

Triplestores or resource description framework (RDF) stores are purpose-built databases used to organise, store and share data with context. Knowledge extraction from a large amount of interconnected data requires effective tools and methods to address the complexity and the underlying structure of semantic information. We propose a method that generates an interpretable multilayered network from an RDF database. The method utilises frequent itemset mining (FIM) of the subjects, predicates and the objects of the RDF data, and automatically extracts informative subsets of the database for the analysis. The results are used to form layers in an analysable multidimensional network. The methodology enables a consistent, transparent, multi-aspect-oriented knowledge extraction from the linked dataset. To demonstrate the usability and effectiveness of the methodology, we analyse how the science of sustainability and climate change are structured using the Microsoft Academic Knowledge Graph. In the case study, the FIM forms networks of disciplines to reveal the significant interdisciplinary science communities in sustainability and climate change. The constructed multilayer network then enables an analysis of the significant disciplines and interdisciplinary scientific areas. To demonstrate the proposed knowledge extraction process, we search for interdisciplinary science communities and then measure and rank their multidisciplinary effects. The analysis identifies discipline similarities, pinpointing the similarity between atmospheric science and meteorology as well as between geomorphology and oceanography. The results confirm that frequent itemset mining provides an informative sampled subsets of RDF databases which can be simultaneously analysed as layers of a multilayer network.

AbstractA vast amount of usable electronic data is in the form of unstructured text. The relation extraction task aims to identify useful information in text (e.g. PersonW works for OrganisationX, GeneY encodes ProteinZ) and recode it in a format such as a relational database or RDF triplestore that can be more effectively used for querying and automated reasoning. A number of resources have been developed for training and evaluating automatic systems for relation extraction in different domains. However, comparative evaluation is impeded by the fact that these corpora use different markup formats and notions of what constitutes a relation. We describe the preparation of corpora for comparative evaluation of relation extraction across domains based on the publicly available ACE 2004, ACE 2005 and BioInfer data sets. We present a common document type using token standoff and including detailed linguistic markup, while maintaining all information in the original annotation. The subsequent reannotation process normalises the two data sets so that they comply with a notion of relation that is intuitive, simple and informed by the semantic web. For the ACE data, we describe an automatic process that automatically converts many relations involving nested, nominal entity mentions to relations involving non-nested, named or pronominal entity mentions. For example, the first entity is mapped from ‘one’ to ‘Amidu Berry’ in the membership relation described in ‘Amidu Berry, one half of PBS’. Moreover, we describe a comparably reannotated version of the BioInfer corpus that flattens nested relations, maps part-whole to part-part relations and maps n-ary to binary relations. Finally, we summarise experiments that compare approaches to generic relation extraction, a knowledge discovery task that uses minimally supervised techniques to achieve maximally portable extractors. These experiments illustrate the utility of the corpora.1

The Finnish Biodiversity Information Facility, FinBIF (https://species.fi), has developed its own taxon database. This allows FinBIF taxon specialists to maintain their own, expert-validated view of Finnish species. The database covers national needs and can be rapidly expanded by our own development team. Furthermore, in the database each taxon is given a globally unique persistent URI identifier (https://www.w3.org/TR/uri-clarification), which refers to the taxon concept, not just to the name. The identifier doesn’t change if the taxon concept doesn’t change. We aim to ensure compatibility with checklists from other countries by linking taxon concepts as Linked Data (https://www.w3.org/wiki/LinkedData) — a work started as a part of the Nordic e-Infrastructure Collaboration (NeIC) DeepDive project (https://neic.no/deepdive). The database is used as a basis for observation/specimen searches, e-Learning and identification tools, and it is browsable by users of the FinBIF portal. The data is accessible to everyone under CC-BY 4.0 license (https://creativecommons.org/licenses/by/4.0) in machine readable formats. The taxon specialists maintain the taxon data using a web application. Currently, there are 60 specialists. All changes made to the data go live every night. The nightly update interval allows the specialists a grace period to make their changes. Allowing the taxon specialists to modify the taxonomy database themselves leads to some challenges. To maintain the integrity of critical data, such as lists of protected species, we have had to limit what the specialists can do. Changes to critical data is carried out by an administrator. The database has special features for linking observations to the taxonomy. These include hidden species aggregates and tools to override how a certain name used in observations is linked to the taxonomy. Misapplied names remain an unresolved problem. The most precise way to record an observation is to use a taxon concept: Most observations are still recorded using plain names, but it is possible for the observer to pick a concept. Also, when data is published in FinBIF from other information systems, the data providers can link their observations to the concepts using the identifiers of concepts. The ability to use taxon concepts as basis of observations means we have to maintain the concepts over time — a task that may become arduous in the future (Fig. 1). As it stands now, the FinBIF taxon data model — including adjacent classes such as publication, person, image, and endangerment assessments — consists of 260 properties. If the data model were stored in a normalized relational database, there would be approximately 56 tables, which could be difficult to maintain. Keeping track of a complete history of data is difficult in relational databases. Alternatively, we could use document storage to store taxon data. However, there are some difficulties associated with document storages: (1) much work is required to implement a system that does small atomic update operations; (2) batch updates modifying multiple documents usually require writing a script; and (3) they are not ideal for doing searches. We use a document storage for observation data, however, because they are well suited for storing large quantities of complex records. In FinBIF, we have decided to use a triplestore for all small datasets, such as taxon data. More specifically, the data is stored according to the RDF specification (https://www.w3.org/RDF). An RDF Schema defines the allowed properties for each class. Our triplestore implementation is an Oracle relational database with two tables (resource and statement), which gives us the ability to do SQL queries and updates. Doing small atomic updates is easy as only a small subset of the triplets can be updated instead of the entire data entity. Maintaining a complete record of history comes without much effort, as it can be done on an individual triplet level. For performance-critical queries, the taxon data is loaded into an Elasticsearch (https://www.elastic.co) search engine.

In times of emerging diseases, data sharing and data integration are of particular relevance for One Health Surveillance (OHS) and decision support. Furthermore, there is an increasing demand to provide governmental data in compliance to the FAIR (Findable, Accessible, Interoperable, Reusable) data principles. Semantic web technologies are key facilitators for providing data interoperability, as they allow explicit annotation of data with their meaning, enabling reuse without loss of the data collection context. Among these, we highlight ontologies as a tool for modeling knowledge in a field, which simplify the interpretation and mapping of datasets in a computer readable medium; and the Resource Description Format (RDF), which allows data to be shared among human and computer agents following this knowledge model. Despite their potential for enabling cross-sectoral interoperability and data linkage, the use and application of these technologies is often hindered by their complexity and the lack of easy-to-use software applications. To overcome these challenges the OHEJP Project ORION developed the Health Surveillance Ontology (HSO). This knowledge model forms a foundation for semantic interoperability in the domain of One Health Surveillance. It provides a solution to add data from the target sectors (public health, animal health and food safety) in compliance with the FAIR principles of findability, accessibility, interoperability, and reusability, supporting interdisciplinary data exchange and usage. To provide use cases and facilitate the accessibility to HSO, we developed the One Health Linked Data Toolbox (OHLDT), which consists of three new and custom-developed web applications with specific functionalities. The first web application allows users to convert surveillance data available in Excel files online into HSO-RDF and vice versa. The web application demonstrates that data provided in well-established data formats can be automatically translated in the linked data format HSO-RDF. The second application is a demonstrator of the usage of HSO-RDF in a HSO triplestore database. In the user interface of this application, the user can select HSO concepts based on which to search and filter among surveillance datasets stored in a HSO triplestore database. The service then provides automatically generated dashboards based on the context of the data. The third web application demonstrates the use of data interoperability in the OHS context by using HSO-RDF to annotate meta-data, and in this way link datasets across sectors. The web application provides a dashboard to compare public data on zoonosis surveillance provided by EFSA and ECDC. The first solution enables linked data production, while the second and third provide examples of linked data consumption, and their value in enabling data interoperability across sectors. All described solutions are based on the open-source software KNIME and are deployed as web service via a KNIME Server hosted at the German Federal Institute for Risk Assessment. The semantic web extension of KNIME, which is based on the Apache Jena Framework, allowed a rapid an easy development within the project. The underlying open source KNIME workflows are freely available and can be easily customized by interested end users. With our applications, we demonstrate that the use of linked data has a great potential strengthening the use of FAIR data in OHS and interdisciplinary data exchange.