Dissertations / Theses on the topic 'Adaptive preferences'

Consider the following scenario: A mother chooses to marry off her 10 year-old daughter, not because she doesn’t know the harmful effects of child marriage, nor because she thinks that it is good that her daughter marries when she is 10 years old. Rather, she is unable to feed her daughter and realizes that her daughter’s survival depends upon her marrying a financially stable man. This is an apparent example of what human development practitioners and political philosophers call an adaptive preference (AP): a preference, formed under oppressive circumstances, that seems to perpetuate the agent’s own oppression. Prevailing opinion is that forced tradeoffs—especially following Serene Khader’s taxonomy—, like the case presented above, are a type of AP: one in which a person makes a decision because of a limited option set. In this paper I argue that no paradigm cases of forced tradeoffs should not be classified as APs. Instead, I offer a revised definition of adaptive preferences where I argue that adaptive preferences are psychological traits that cause the agent with adaptive preferences to make irrational or uninformed decisions that perpetuate their own oppression. I defend this new definition by exploring the implications of changing the definition. In particular, forced tradeoffs involve different kinds of interventions from other kinds of adaptive preferences and including forced tradeoffs risks committing testimonial injustice against those who have limited option sets.
Master of Arts

In comparison to conventional cruise control, adaptive cruise control (ACC) vehicles are capable of sensing forward traffic and slowing to accommodate as necessary. When no forward vehicles are present, ACC function is the same as conventional cruise control. However, with ACC, when a slower vehicle is detected, the ACC system will decelerate and follow at a selected time-based distance. While slowing to follow, the driver will experience a system-controlled deceleration of the ACC vehicle. An experiment was conducted to evaluate driver preferences for the distance at which the primary deceleration occurs and the level of deceleration that is obtained. Driver intervention was required in one trial and driver response behavior was measured. Ten men and ten women in two age groups evaluated the decelerations from a cruise speed of 70mph to a following speed of 55mph behind a confederate lead vehicle on the highway. Evaluations can be made using four scales: Good vs. Bad, Comfortable vs. Uncomfortable, Jerky vs. Smooth, and Early vs. Late. Decelerations of approximately 0.06g which occur approximately 200ft to 250ft behind the lead vehicle were most preferred. Prior to intervention, foot position ranged from a point directly below the brake pedal to 16.4in from the brake pedal. Foot motion began between 21.12s time-to-collision (TTC) and 3.97s TTC. Eighty percent of the participants paused to "cover" the brake before final motion to activate the brake. The older age group intervened (braked) later than the younger age group. Driver braking after intervention ranged from 0.16g to 0.32g.
Master of Science

This study attempted to combine the benefits of multimedia learning, adaptive interfaces, and learning style theory by constructing a novel e-learning environment. The environment was designed to accommodate individual learning styles while students progressed through a computer programming course. Despite the benefits of individualised instruction and a growing worldwide e-learning market, there is a paucity of guidance on how to effectively accommodate learning styles in an online environment. Several existing learning-style adaptive environments base their behaviour on an initial assessment of the learner's profile, which is then assumed to remain stable. Consequently, these environments rarely offer the learner choices between different versions of content. However, these choices could cater for flexible learning styles, promote cognitive flexibility, and increase learner control. The first research question underlying the project asked how learning styles could be accommodated in an adaptive e-learning environment. The second question asked whether a dynamically adaptive environment that provides the learner with a choice of media experiences is more beneficial than a statically adapted environment. To answer these questions, an adaptive e-learning environment named iWeaver was created and experimentally evaluated. iWeaver was based on an introductory course in Java programming and offered learning content as style-specific media experiences, assisted by additional learning tools. These experiences and tools were based on the perceptual and information processing dimension of an adapted version of the Dunn and Dunn learning styles model. An experimental evaluation of iWeaver was conducted with 63 multimedia students. The analysis investigated the effect of having a choice of multiple media experiences (compared to having just one static media experience) on learning gain, enjoyment, perceived progress, and motivation. In addition to these quantitative measurements, learners provided qualitative feedback at the end of each lesson. Data from 27 participants were sufficiently complete to be analysed. For the data analysis, participants were divided into two groups of high and low interest in programming and Java, then into two groups of high and low experience with computers and the Internet. Both group comparisons revealed statistically significant differences for the effect of choice. Having a choice of media experiences proved beneficial for learners with low experience but detrimental for learners with high experience or interest. These findings suggest that the effect of choice appears to be strongly influenced by the learner's background. It is hypothesised that encouraging a more active learner role in educational systems would expand the positive influence of choice to a wider range of learners. The study has contributed some weight to the argument that for certain groups of learners, it is more beneficial to view learning style as a flexible, rather than a stable construct. As a practical implication, it seems advisable to collect data on prior experience, interest, and the initial learning style distribution of the target audience before developing environments comparable to iWeaver. [See http://www.adaptive-learning.net/research/media.htm for media files associated with this thesis.]

All’interno del quadro teorico del Capability Approach ‘agency’ viene definito come ciò che permette agli individui di raggiungere i propri obiettivi. Quando il concetto di ‘agency’ viene studiato attraverso misure soggettive, insorge il problema delle preferenze adattive. Lo scopo di questa Tesi è di testare una metodologia che possa catturare la percezione individuale di ‘agency’ nelle diverse dimensioni che compongono la nostra vita, limitando i problemi di misurazione dovuti delle preferenze adattive. Per questo fine, ci affidiamo al ‘Relative Autonomy Index’. Inoltre, utilizziamo la metodologia di ‘Achoring Vignette’ per aumentare la comparabilità tra le percezioni dei singoli individui. Nella prima parte, indaghiamo se il nostro metodo effettivamente riduce l’incomparabilità individuale attraverso uno studio pilota. I risultati dimostrano come il nostro metodo sia in grado di catturare differenze nel comportamento di risposta. Nella seconda parte, sulla base di una seconda raccolta di dati, applichiamo la nostra metodologia al fine di indagare l'impatto dei Gruppi di Acquisto Solidale sul livello di autonomia di un gruppo di produttori italiani. I nostri risultati indicano che il coinvolgimento con i Gruppi di Acquisto Solidale evita la riduzione dell'autonomia percepita dai quei produttori che non sono coinvolti con i Gruppi di Acquisto Solidale.
Within the Capability Approach, agency is one of the components that enable individuals to be empowered. When subjective measures are applied to the study of agency, the problem of ‘adaptive preferences’ arises. The aim of this thesis is to test a methodology that captures the individual perception of agency, while taking into account the bias coming from adaptive preferences. We rely on the Relative Autonomy Index to capture the individual’s agency in several dimensions. In addition, in order to increase the comparability between the individual’s perceptions, we employ the Anchoring Vignette methodology. We investigate whether our methodology reduces the individual incomparability through a pilot study. The results of this preliminary exercises show that our methodology is able to capture dissimilarities in response behaviours. Through a second data collection, we apply our methodology in order to investigate the impact that the commercial relationship between Solidarity Purchasing Group and Italian producers has on the latter. Our findings suggest that the involvement in the Solidarity Purchasing Group has a mainly positive impact on producers’ level of autonomy, which remains at lower levels for producers that did not develop a relationship with the Solidarity Purchasing Group.

All’interno del quadro teorico del Capability Approach ‘agency’ viene definito come ciò che permette agli individui di raggiungere i propri obiettivi. Quando il concetto di ‘agency’ viene studiato attraverso misure soggettive, insorge il problema delle preferenze adattive. Lo scopo di questa Tesi è di testare una metodologia che possa catturare la percezione individuale di ‘agency’ nelle diverse dimensioni che compongono la nostra vita, limitando i problemi di misurazione dovuti delle preferenze adattive. Per questo fine, ci affidiamo al ‘Relative Autonomy Index’. Inoltre, utilizziamo la metodologia di ‘Achoring Vignette’ per aumentare la comparabilità tra le percezioni dei singoli individui. Nella prima parte, indaghiamo se il nostro metodo effettivamente riduce l’incomparabilità individuale attraverso uno studio pilota. I risultati dimostrano come il nostro metodo sia in grado di catturare differenze nel comportamento di risposta. Nella seconda parte, sulla base di una seconda raccolta di dati, applichiamo la nostra metodologia al fine di indagare l'impatto dei Gruppi di Acquisto Solidale sul livello di autonomia di un gruppo di produttori italiani. I nostri risultati indicano che il coinvolgimento con i Gruppi di Acquisto Solidale evita la riduzione dell'autonomia percepita dai quei produttori che non sono coinvolti con i Gruppi di Acquisto Solidale.
Within the Capability Approach, agency is one of the components that enable individuals to be empowered. When subjective measures are applied to the study of agency, the problem of ‘adaptive preferences’ arises. The aim of this thesis is to test a methodology that captures the individual perception of agency, while taking into account the bias coming from adaptive preferences. We rely on the Relative Autonomy Index to capture the individual’s agency in several dimensions. In addition, in order to increase the comparability between the individual’s perceptions, we employ the Anchoring Vignette methodology. We investigate whether our methodology reduces the individual incomparability through a pilot study. The results of this preliminary exercises show that our methodology is able to capture dissimilarities in response behaviours. Through a second data collection, we apply our methodology in order to investigate the impact that the commercial relationship between Solidarity Purchasing Group and Italian producers has on the latter. Our findings suggest that the involvement in the Solidarity Purchasing Group has a mainly positive impact on producers’ level of autonomy, which remains at lower levels for producers that did not develop a relationship with the Solidarity Purchasing Group.

Adaptive co-management (ACM) has downplayed the role that diversity withincommunities could play in management schemes. It has understood communities asgroups of persons with similar interests. This thesis attempts to explore some of theconsequences that diversity of preferences of the future over social-ecologicaltrajectories within communities could have on adaptive co-management processes.The relevance of understanding this lies in the urgency of finding solutions thatpromote better resource and environmental management aimed at sustainability. Thefollowing work uses scenarios as a way to explore the effects that diversity ofpreferences within communities might have on ACM processes.

Today's mobile computing devices provide a convenient means to search for points-of-interest (POIs) such as restaurants and accommodation. Mobile Preference-Based Search Tools (PBSTs) allow users to identify POIs such as restaurants or accommodation most suited to their needs and constraints using a mobile device. These devices however, have several design constraints including limited screen space and hardware capabilities. Adaptive User Interfaces (AUIs) have been proposed to address these issues but have not been extensively applied to mobile PBSTs such as mobile tourist guides. In addition, AUIs possess several benefits and advantages over static (traditional) interfaces, which do not take a user's preferences, skill set and experience into account. Little research, however, has been conducted into identifying the potential benefits of AUIs for mobile preference-based searching (PBS). The aim of this research was to determine the extent to which an AUI could improve the effectiveness and user satisfaction of mobile PBS. A literature study was conducted to determine the benefits and limitations of existing mobile PBSTs and determine how these could be improved. The potential benefits of AUIs for mobile PBSTs and a mobile map-based visualisation system were identified. A suitable model for incorporating an AUI into a mobile PBST was identified. The requirements for a mobile PBST were combined with the potentially adaptable objects of a Mobile Map-based Visualisation (MMV) system to provide adaptation suggestions for POInter, an existing mobile tourist guide. A field study using POInter was conducted in order to measure the extent to which participants agreed with suggestions provided for adapting the information, interaction and visualisation aspects of the system. These results were used to derive adaptation requirements for A-POInter, an adaptive version of POInter. Using a model-based design approach, an AUI was designed and implemented for A-POInter. An extensive field study was then conducted to evaluate the usability of the adaptations provided by A-POInter. The quantitative and qualitative data collected from the evaluations allowed the usability of A-POInter to be determined. The results of the field study showed that the participants were highly satisfied with the usability and the usefulness of the adaptations provided by A-POInter. Conclusions and recommendations for future work based on the results of the research were then outlined to conclude the dissertation.

First lines of the Introduction (as abstract not provided): This thesis seeks to primarily answer the following question: are adapted preferences autonomous? In pursuing the answer of this question, I am unsurprisingly faced with two importantly related queries: firstly, what actually is adaptive preference formation? And secondly, what kind of theory of autonomy is correct and why? In the spirit of question answering, the first chapter of this thesis seeks to provide a more robust account of adaptive preference formation (herein APF), a theory which states that the preferences held by an agent can be subconsciously causally produced by the restriction of options. Through an examination of Jon Elster’s original account of the concept, and a consideration of Amartya Sen and Martha Nussbaum’s contemporary interpretations of Elster’s account, I intend to flesh out the mechanics of APF, considering the necessary and sufficient conditions for APF. This section aims to solidify the descriptive literature of APF, with a focus on differentiating the process from other similar concepts such as character planning and internalised oppression (herein IO). Ultimately, I conclude with a variation of Elster’s account and produce my own examples of agents who hold adapted preferences (herein AP).

Oramai, al giorno d'oggi la gente si connette ad Internet attraverso una moltitudine di Smart device differenti. La riproduzione di materiale video multimediale è l'attività maggiormente richiesta durante la navigazione in Internet. Essa sfrutta gran parte della larghezza di banda della rete a disposizione, con una tendenza che molto probabilmente nel futuro non farà altro che aumentare. Per contrastare la variabilità delle risorse di internet, gli algoritmi di rate adaptation vengono usati per adattare la velocità di trasmissione alle capacità del network. In questo elaborato andremo a presentare un nuovo algoritmo di rate adaptation per l'adaptive HTTP streaming che vuole dare la possibilità all'utente di avere una riproduzione fluida, caratterizzata dal minore numero di pause, ma che allo stesso tempo sia in grado di garantire una alta qualità della riproduzione del materiale multimediale. Dopo una analisi iniziale dell'algoritmo andremo ad analizzare i benefici della nostra proposta in termini di qualità di riproduzione, numero di pause, tempo di pause, ecc.., rispetto ad altri algoritmi di rate adaptation. I nostri risultati dimostrano come l'algoritmo, adattandosi dinamicamente allo stato della connessione riesca a mantenere un basso numero di pause mentre aumenta la qualità dello streaming.

Robotic assistants have inspired numerous books and science fiction movies. In the real world, these kinds of devices are a growing need in amongst the elderly, who while life continue requiring more assistance. While life expectancy is increasing, life quality is not necessarily doing so. Thus, we may find ourselves and our loved ones being dependent and needing another person to perform the most basic tasks, which has a strong psychological impact. Accordingly, assistive robots may be the definitive tool to give more quality of life by empowering dependent people and extending their independent living. Assisting users to perform daily activities requires adapting to them and their needs, as they might not be able to adapt to the robot. This thesis tackles adaptation and personalization issues through user preferences. We 'focus on physical tasks that involve close contact, as these present interesting challenges, and are of great importance for he user. Therefore, three tasks are mainly used throughout the thesis: assistive feeding, shoe fitting, and jacket dressing. We first describe a framework for robot behavior adaptation that illustrates how robots should be personalized for and by end- users or their assistants. Using this framework, non-technical users determine how !he robot should behave. Then, we define the concept of preference for assistive robotics scenarios and establish a taxonomy, which includes hierarchies and groups of preferences, grounding definitions and concepts. We then show how the preferences in the taxonomy are used with Al planning systems to adapt the robot behavior to the preferences of the user obtained from simple questions. Our algorithms allow for long-term adaptations as well as to cope with misinformed user models. We further integrate the methods with low-level motion primitives that provide a more robust adaptation and behavior while lowering the number of needed actions and demonstrations. Moreover, we perform a deeper analysis in Planning and preferences with the introduction of new algorithms to provide preference suggestions in planning domains. The thesis then concludes with a user study that evaluates the use of the preferences in the three real assistive robotics scenarios. The experiments show a clear understanding of the preferences of users, who were able to assess the impact of their preferences on the behavior of the robot. In summary, we provide tools and algorithms to design the robotic assistants of the future. Assistants that should be able to adapt to the assisted user needs and preferences, just as human assistants do nowadays.
Els assistents robòtics han inspirat nombrosos llibres i pel·lícules de ciència-ficció al llarg de la història. Però tornant al món real, aquest tipus de dispositius s'estan tornant una necessitat per a una societat que envelleix a un ritme ràpid i que, per tant, requerirà més i més assistència. Mentre l'esperança de vida augmenta, la qualitat de vida no necessàriament ho fa. Per tant, ens podem trobar a nosaltres mateixos i als nostres estimats en una situació de dependència, necessitant una altra persona per poder fer les tasques més bàsiques, cosa que té un gran impacte psicològic. En conseqüència, els robots assistencials poden ser l'eina definitiva per proporcionar una millor qualitat de vida empoderant els usuaris i allargant la seva capacitat de viure independentment. L'assistència a persones per realitzar tasques diàries requereix adaptar-se a elles i les seves necessitats, donat que aquests usuaris no poden adaptar-se al robot. En aquesta tesi, abordem el problema de l'adaptació i la personalització d'un robot mitjançant preferències de l'usuari. Ens centrem en tasques físiques, que involucren contacte amb la persona, per les seves dificultats i importància per a l'usuari. Per aquest motiu, la tesi utilitzarà principalment tres tasques com a exemple: donar menjar, posar una sabata i vestir una jaqueta. Comencem definint un marc (framework) per a la personalització del comportament del robot que defineix com s'han de personalitzar els robots per usuaris i pels seus assistents. Amb aquest marc, usuaris sense coneixements tècnics són capaços de definir com s'ha de comportar el robot. Posteriorment definim el concepte de preferència per a robots assistencials i establim una taxonomia que inclou jerarquies i grups de preferències, els quals fonamenten les definicions i conceptes. Després mostrem com les preferències de la taxonomia s'utilitzen amb sistemes planificadors amb IA per adaptar el comportament del robot a les preferències de l'usuari, que s'obtenen mitjançant preguntes simples. Els nostres algorismes permeten l'adaptació a llarg termini, així com fer front a models d'usuari mal inferits. Aquests mètodes són integrats amb primitives a baix nivell que proporcionen una adaptació i comportament més robusts a la mateixa vegada que disminueixen el nombre d'accions i demostracions necessàries. També fem una anàlisi més profunda de l'ús de les preferències amb planificadors amb la introducció de nous algorismes per fer suggeriments de preferències en dominis de planificació. La tesi conclou amb un estudi amb usuaris que avalua l'ús de les preferències en les tres tasques assistencials. Els experiments demostren un clar enteniment de les preferències per part dels usuaris, que van ser capaços de discernir quan les seves preferències eren utilitzades. En resum, proporcionem eines i algorismes per dissenyar els assistents robòtics del futur. Uns assistents que haurien de ser capaços d'adaptar-se a les preferències i necessitats de l'usuari que assisteixen, tal com els assistents humans fan avui en dia.

Some women in India are making the decision to abort their foetus because it has been identified as female. One feminist intuition could be that those decisions are a result of false consciousness, that women have adapted their preferences in response to the social and gender oppression to which they are subjected. The presumption is also that the adaptive preference signifies an absence of autonomy. This thesis challenges that intuition and presumption, arguing that not all preferences made under oppressive conditions are to be regarded as non-autonomous. The motivation behind decisions is complex and a decision must be considered in the context of varied social, economic and cultural influences. The thesis uses three hypothetical case studies to facilitate a critical analysis of the concept of adaptive preference formation and different theories of autonomy from a feminist perspective: the content-neutral traditional procedural theory, content laden substantive theories and the relational theories which are re-conceptualised versions of the former accounts. This results in a theoretical analysis of which approach to autonomy best acknowledges the effect oppression may have on decision-making but is also able to recognise a woman’s autonomy where possible. This analysis is valuable as it reveals the difficulties in determining autonomy when such complex theories of autonomy are applied to real-life cases but also demonstrates that there is the possibility of the presence of autonomy even within a restricted choice environment. Although this focus on autonomy must not serve to detract from condemnation of a practice which perpetuates the denigration of women, condemnation alone is unlikely to result in a change in practice. I conclude that only when respect for bodily integrity is placed at the heart of medical decision-making in India will the right to self-determination and autonomy be realised for everyone.

This thesis examines the ontogeny of plasma sodium regulation (an indicator of osmoregulatory ability), current or rheotactic response (an indicator of emigration timing) and salinity preference in juvenile coho salmon (Oncorhynchus kisutch). The purpose of the study was to determine if there are inherited differences in the development of these traits between coastal and interior British Columbia populations of coho. An interior (Cold water River) and a coastal (Rosewall Creek-Big Qualicum River) population were monitored for the above traits throughout the year. Both wild and laboratory groups were included in the study. The laboratory raised populations were divided into two incubation treatment groups: one incubated under a coastal temperature regime, and the other incubated under an interior temperature regime. There were no differences in the development of sodium regulatory ability between wild populations when the data were sorted by coho weight. Coastal coho, however, physiologically smolted after one year in the natal streams, while interior coho smolted after at least two years of freshwater growth. No obvious differences were noted between wild resident populations in the timing of downstream movement or the shift in salinity preference from hypotonic to isotonic and hypertonic salinities. Both of these behavioural responses typically occurred in the spring (April-May) of each year. Fyke net catches, however, sugqested that, in addition to the spring emigrations observed in both populations, a portion of the interior population migrated in the fall (November). No differences in the development of sodium regulatory ability were observed either within or between laboratory raised populations. Ion regulatory ability increased to a plateau in the fall and winter following emergence, and increased to smolting levels during the following spring (April-May). There were differences between coastal and interior populations in the pattern of development of both nocturnal current responses and the preference for isotonic or hypertonic salinities. Interior laboratory raised coho developed negative nocturnal rheotaxis and a preference for isotonic salinities about three months earlier (November) than laboratory raised coastal coho (late February-March). Within populations, no differences were observed in the ontogeny of these traits in the groups reared under different temperature regimes. Because these interpopulation ontogenetic behavioural differences persisted in fish reared under identical laboratory conditions, they probably have some genetic basis. Such an innate component in behaviour implies an adaptive role and in juvenile coho these behavioural traits may allow populations to use a variety of habitats at different distances from the sea, even though a major physiological schedule (in this case the development of ion regulatory capabilities) appears to be fixed within the species. Perhaps variations in migratory timing and salinity preference in juvenile coho evolved to assure survival in a relatively unstable and often severe environment by optimizing habitat use within the constraints of an overriding physiological schedule.
Science, Faculty of
Zoology, Department of
Graduate

What should we think of the woman who believes that it is proper for her husband to make all of the family's decisions? How do we respond to the mother who thinks that her domestic abuse is deserved? One popular explanation of such cases is that people's preferences may "adapt" to harmful or unacceptable circumstances that they see as unavoidable, so that they come to endorse states of affairs that they should rightly resist. Yet the concept of adaptive preferences is highly controversial: while its defenders argue that it is a useful tool for understanding the way in which members of marginalized groups can acquiesce to their own oppression and marginalization, its detractors argue that employing the concept unavoidably and disrespectfully treats adults as if they were children who do not know what is good for them. But at the same time, the concept is incredibly under-theorized: defenders and detractors of adaptive preferences alike rarely offer a clear account of the concept against which their claims can be tested. In this thesis, I propose and defend an autonomy-based account of adaptive preferences with four aims: a) conceptual clarity, b) the ability to show respect for persons, d) conduciveness to the political project of reducing marginalization, and d) recognition of and attention to the differential needs of children and adults. To achieve the first aim, I outline what I call an ""indirect substantive"" account of autonomy. This account uses substantive content indirectly to determine whether a person's preferences count as autonomous. Because my account of autonomy uses substantive content indirectly, it allows the possibility that any preference could count as autonomous, and therefore non-adaptive. In this way, it shows respect for persons by recognizing the role that their own processes of moral reasoning play in determining their good. But by using substantive content, I also ensure that the concept can aid the political project. In order for a preference to count as autonomous, and therefore non-adaptive, I argue that the person developing it must have been exposed to alternatives to the preference developed that were both live and valuable. In this way, while people may legitimately prefer options that seem to third parties to be unattractive or marginalizing, these preferences only deserve full deference when they have been chosen from among valuable alternatives. Finally, I turn to the issue of children. This issue is especially important, since previous discussions of adaptive preferences have failed to recognize that the needs of adults who have already developed adaptive preferences are very different from the needs of children in the process of developing them. I argue for ways of operationalizing my theoretical account of adaptive preferences that can both show respect for the already-formed preferences of adults (including those that are adaptive), and prevent children from forming adaptive preferences in the first place.

Web-based learning environments often use games and simulations to enrich the learning process. Understanding the response of learners to these non-text-based environments is usually accomplished through laboratory testing or field surveys. One common method of gathering user feedback is the "click-to-submit" web-form questionnaire. However, these questionnaires are plagued by low response rates and inconsistent results. Research into survey design reveals many potential problems in the construction and administration of questionnaires — problems that are exacerbated by the nature of communication over the web. With its interactive NFBkids website, the National Film Board of Canada uses web-form questionnaires to gather user feedback. Unfortunately, the web-based questionnaires at NFBkids fail to obtain useful feedback from those who visit the website. Moreover, the managers of NFBkids feel the website fails to create a sense of "community" among the site's users — one of the design goals of NFBkids. By applying existing research into cognitive theory and psycholinguistics to the design of interactive methods of gathering user feedback, this paper investigates an alternative methodology for the collection of user preferences. In addition to the familiar "clickable" web-form, the research described in this paper explores interactive tools that complement web-forms as a method for gathering feedback from users. Real-time, peer-to-peer assessment techniques are introduced as a way to collect valuable user feedback and to foster a sense of peer presence among a community of web users.

碩士
中華大學
資訊工程學系碩士班
87
Abstract With the explosive growth of on-line information on various kinds of platforms (e.g. the library and the Internet), users may have more opportunities to get information conveniently. However, without an efficient and personalized information retrieval (IR) system, the users could also be misleader in the huge amount of information. Therefore, how to search for useful information for each individual users has become an urgent problem for most researchers in IR, library science and the Internet. To attack the problem, IR systems should be able to consider different users' information needs. In this thesis, I propose a model for building adaptive information retrieval (AIR) systems which may adapt its search strategies to users' different information needs and preferences. Thus, both the quality and the efficiency of IR may be promoted. In particular, AIR should be able to (1) recognize the personal information needs of individual users, (2) detect the change of user preferences, and (3) map the user’s semantic preferences to the contents of information. However, according to the survey conducted in the thesis, all search engines on the Internet are unable to achieve the three tasks. Because users' information needs and preferences are often expressed in their queries, AIR should observe and acquire users' semantic preferences from their queries. The output of the learning module is the mapping between query terms and their suitable subset of the document database. As next query is entered, its terms are extracted and mapped to suitable document database in which useful information is more likely to be found. AIR employs a set of the Heuristic Semantic Patterns (HSPs) to learn users' semantic preferences. The HSPs works on a tree-structured document database which is common for most libraries and web sites on the Internet. A semantic preference of a term is expressed as a mapping between the term and its suitable document category on the document database. As more semantic preferences of query terms may be acquired, AIR may adapt its search strategy to individual needs and preferences. In this thesis, AIR is explored both theoretically and empirically. The impacts and contributions of the work will be evaluated in terms of the extent to which both the quality and the efficiency of the IR are improved.

Tese de doutoramento em Engenharia Eletrotécnica, na especialidade de Otimização e Teoria de Sistemas, apresentada à Faculdade de Ciências e Tecnologia da Universidade de Coimbra.
The resolution of a multi-objective optimization problem involves, in general, not only a search phase adequate to provide a representative set of the Pareto-optimal front, but also a decision phase consisting in the identification of a solution (or a set of solutions) acceptable as a final recommendation having in mind practical implementation. The incorporation of preferences during the evolutionary process allows focusing the search according to the preference information elicited from the decision maker, avoiding the exploration of irrelevant solutions (thus minimizing the computational time) and facilitating the integration of knowledge in the search process (minimizing the cognitive effort). These aspects are particularly important in combinatorial problems, when the number of objective functions is large and/or their nature is conflicting, since the size of the search space as well as the number of non-dominated solutions tends to be very high. The evolutionary approach, called EvABOR (Evolutionary Algorithm Based on an outranking Relation), presented in this work incorporates the decision maker’s preferences to guide the search for regions of the space more in accordance with the elicited preferences. These are captured and made operational using the principles and parameters of the ELECTRE TRI method. The outranking relation in the ELECTRE TRI method is used to replace/complement the non-dominance relation in the usual evolutionary algorithm operators (crossover, mutation and selection). Since the quality of the initial solutions may influence the performance of an evolutionary algorithm a methodology based on GRASP (Greedy Randomized Adaptive Search Procedure) is proposed for the construction of initial solutions. This procedure is particularly relevant when knowledge about the problem at hand exists, which happens, in general, in real-world problems. Additionally, the need to exploit regions of the search space more efficiently led to the implementation of a local search procedure based on Simulated Annealing, in which the preferences elicited from a decision maker are taken into account. This motivated the development of a new approach for multi-objective optimization problems in which GRASP and Simulated Annealing are hybridized, incorporating preferences in the construction phase or/and the local search phase. The proposed algorithms are applied to provide decision support in the resolution of two real-world problems: a reactive power compensation problem in electrical distribution networks, using the EvABOR algorithm, and a direct load control problem, using the hybrid algorithm.
A resolução de um problema de optimização multiobjectivo envolve, em geral, não apenas uma fase de pesquisa, capaz de fornecer um conjunto representativo da frente óptima de Pareto, mas também uma fase de decisão, consistindo na escolha da solução (ou conjunto de soluções) aceitável como recomendação final tendo em vista a sua aplicação prática. Neste sentido, a incorporação de preferências durante o processo evolutivo permite focar a pesquisa evitando a exploração de soluções irrelevantes (minimizando assim o tempo de computação) e facilita a integração de conhecimento do decisor no processo de pesquisa (minimizando o esforço cognitivo). Estes aspectos são particularmente importantes quando o número de funções objectivo é grande e/ou a sua natureza é conflituante, uma vez que a dimensão do espaço de pesquisa assim como o número de soluções não-dominadas admissíveis tende a ser elevado. A proposta de uma abordagem evolutiva, designada por EvABOR (Evolutionary Algorithm Based on an Outranking Relation), apresentada neste trabalho incorpora as preferências de um decisor de modo a guiar a pesquisa para regiões do espaço mais de acordo com as preferências explicitadas. Estas são captadas e tornadas operacionais recorrendo aos parâmetros e princípios do método ELECTRE TRI. A relação de prevalência (outranking), na qual o ELECTRE TRI se baseia, é usada para substituir/complementar a relação de não dominância nos habituais operadores do algoritmo evolutivo (cruzamento, mutação e selecção). Dado que a qualidade das soluções iniciais pode influenciar o desempenho de um algoritmo evolutivo, e existindo conhecimento sobre o problema em causa, nomeadamente ao lidar com problemas reais, propõe-se uma metodologia de construção de soluções iniciais baseada no GRASP (Greedy Randomized Adaptive Search Procedure) permitindo também a incorporação de preferências. Adicionalmente, a necessidade de explorar as regiões do espaço de pesquisa de forma mais eficiente, levou à implementação de um procedimento de pesquisa local, baseado no Simulated Annealing, onde as preferências explicitadas pelo decisor são tidas em conta, sendo também incorporadas numa versão multiobjectivo do Simulated Annealing. Este trabalho teve como resultado um novo algoritmo onde se explora a hibridização do GRASP com o Simulated Annealing, incorporando as preferências tanto na fase de construção como na fase de pesquisa local. Os algoritmos propostos foram aplicados na resolução de dois problemas recorrendo a dados reais: um problema de compensação de energia reactiva em redes de distribuição de energia eléctrica, no caso do EvABOR, e um problema de controlo remoto de cargas, no caso do algoritmo híbrido.

碩士
國立中山大學
資訊管理學系研究所
93
The electronic mail has become one of the most popular communication channels in the modern world. Due to its convenience and low cost, however, many business salesmen utilize this channel to promote their products by distributing e-mails to people as far as they can reach, which causes troubles to irrelevant e-mail receivers. As a result, many a research has been devoted to filtering irrelevant e-mails based on data mining techniques to alleviate users’ mental loadings in processing e-mails they receive. Nevertheless, current approaches have their own drawbacks. Issues on what appropriate classifies to construct, how to endow such classifiers with the adaptive learning ability, and how to customize the e-mail management process for each user are still under investigation. The objective of this research is therefore to construct an e-mail classifier with learning ability to self-correct from erroneous outcomes. Furthermore, we propose a customized e-mail management process that can handle users’ e-mails based on their own preferences. Ultimately, it can adapt itself to the changes of users’ preferences when handling their e-mails. Several experiments are conducted to verify the performance of the constructed classifier. The results show that our proposed classifier possesses high accuracy and high precision with outstanding adaptive learning ability. We also illustrate a real application of the customized e-mail management process. It shows that our approach can detect the changes of users’ preferences and learn to follow the changes. The feasibility of employing our approach to constructing e-mail classifiers is thus justified.

Farmers are increasingly being called upon to help manage, invest and steer water systems towards a desirable state: farmers as water managers. Through on-farm soil and water management and investment decisions, farmers influence their own water availability but their local interventions also have system-level implications. Farmers influence water systems and are in turn influenced by the water system in which they operate. System-level implications of farmers as water managers are poorly understood. This thesis explores approaches and provides insights for a better understanding of the ways in which farmers can contribute to achieving system-level objectives, such as agricultural intensification and freshwater retention capacity. This thesis starts by examining main challenges in assessing the regional impacts of local water storage. By systematically reviewing literature on local water storage, Chapter 2 identifies and discusses technical and socio-economic difficulties encountered in assessing the regional impacts of local interventions. It concludes that the focus of assessments must shift from storage ‘potential’ to storage ‘feasibility’. Feasibility is context specific and influenced by the spatial and temporal scales of analysis. The chapter then further explores farmers’ prefercens and personal motivations for investing (or not investing) in additional water for irrigation. Chapters 3 and 4 present, apply, and evaluate a new framework that uses ‘crossover points’ to support dialogue on irrigation investments in case studies in Tasmania, Australia. The framework extends the use of crossover points in a novel way to facilitate dialogue in a participatory setting, termed ‘participatory crossover analysis’. Participatory crossover analysis proved to perform well as a tool for valuation of irrigation water and to foster social learning. Chapters 3 and 4 investigate farmers’ personal and evolving perspectives on a) their water demand; b) the value of a reliable source of high quality water; and c) their willingness to pay for water. Their personal preferences and reasonings turned out to be diverse and broader than just short-term economic gains. Lifestyle choices, long-term intergenerational planning, perceived risks, and intrinsic motivations were mentioned as factors influencing investment decisions. This has strong implications for the type of information that farmers considered relevant in supporting their decisions on water investments. In short, information and knowledge exchange was highly valued, particularly learning from and with peers. Chapter 5 presents an assessment of social learning during a valuation workshop, using participatory crossover analysis as a tool to facilitate a deliberative dialogue between irrigators, scheme managers, and policymakers about the past, present, and future value of irrigation water. In the case under study, discussions between workshop participants led to new insights on the value of water, identification of potential improvements in management and governance, and cultivated a greater appreciation of the diverse perspectives in the room. These findings suggest that a single workshop can foster social learning. Findings from the Tasmanian cases highlight that the rollout of new irrigation infrastructure triggers social change that is currently not accounted for in the design and management of irrigation schemes. New irrigation schemes are built to operate in a future that cannot be predicted. Conclusions from the cases suggest that management of water systems should be approached as an ongoing process of social learning with stakeholders. Chapter 6 offers a way forward, suggesting an approach to irrigation infrastructure that links insights from the literature on human-water interactions with insights on adaptive pathways. Adaptive approaches to water management better allow farmers to be water managers, today and in the unfolding future.

In perimeter building zones with glass façades, controllable fenestration (daylighting/shading) and electric lighting systems are used as comfort delivery systems under dynamic weather conditions, and their operation affects daylight provision, outside view, lighting energy use, as well as overall occupant satisfaction with the visual environment. A well-designed daylighting and lighting control should be able to achieve high level of satisfaction while minimizing lighting energy consumption. Existing daylighting control studies focus on minimizing energy use with general visual comfort constraints, when adaptive and personalized controls are needed in high performance office buildings. Therefore, reliable and efficient models and methods for learning occupants’ personalized visual preference or satisfaction are required, and the development of optimal daylighting controls requires integrated considerations of visual preference/satisfaction and energy use.

In this Dissertation, a novel method is presented first for developing personalized visual satisfaction profiles in daylit offices using Bayesian inference. Unlike previous studies based on action data, a set of experiments with human subjects was designed and conducted to collect comparative visual preference data (by changing visual conditions) in private offices. A probit model structure was adopted to connect the comparative preference with a latent satisfaction utility model, assumed in the form of a parametrized Gaussian bell function. The distinct visual satisfaction models were then inferred using Bayesian approach with preference data. The posterior estimations of model parameters, and inferred satisfaction utility functions were investigated and compared, with results reflecting the different overall visual preference characteristics discovered for each person.

Second, we present an online visual preference elicitation learning framework for efficiently learning and eliciting occupants’ visual preference profiles and hidden satisfaction utilities. Another set of experiments with human subjects was conducted to implement the proposed learning algorithm in order to validate the feasibility of the method. A combination of Thompson sampling and pure exploration (uncertainty learning) methods was used to balance exploration and exploitation when targeting the near-maximum area of utility during the learning process. Distinctive visual preference profiles of 13 subjects were learned under different weather conditions, demonstrating the feasibility of the learning framework. Entropy of the distribution of the most preferred visual condition is computed for each learned preference profile to quantify the certainty. Learning speed varies with subjects, but using a single variable model (vertical illuminance on the eye), most subjects could be learned to an acceptable certainty level within one day of stable weather, which shows the efficiency of the method (learning outcomes).

Finally, a personalized shading control framework is developed to maximize occupant satisfaction while minimizing lighting energy use in daylit offices with roller shades. An integrated lighting-daylighting simulation model is used to predict lighting energy use while it also provides inputs for computing personalized visual preference profiles, previously developed using Bayesian inference from comparative preference data. The satisfaction utility and the predicted lighting energy use are then used to form an optimization framework. We demonstrate the results of: (i) a single objective formulation, where the satisfaction utility is simply used as a constraint to when minimizing lighting energy use and (ii) a multi-objective optimization scheme, where the satisfaction utility and predicted lighting energy use are formulated as parallel objectives. Unlike previous studies, we present a novel way to apply the MOO without assigning arbitrary weights to objectives: allowing occupants to be the final decision makers in real-time balancing between their personalized visual satisfaction and energy use considerations, within dynamic hidden optimal bounds – through a simple interface.

In summary, we present the first method to incorporate personalized visual preferences in optimal daylighting control, with energy use considerations, without using generic occupant behavior models or discomfort-based assumptions.

碩士
國立臺灣大學
電機工程學研究所
102
This thesis proposes a recommendation system (RS) which incorporates the advantages of the user/item-based collaborative filtering (CF) and the content-based filtering. Unlike the user/item-based CF where the user/item spaces are of high dimension, the proposed RS utilizes the user-based and item-based concept spaces where dimension, or the number of concepts, is increased only necessary. In addition, the proposed system can deal with the cold start problem with producing another kind dimension of items. With modifying clustering results, it can be used to create recommendation in the rapid increasing information. The dimension of the item-based concepts is defined by the features of the items, and concepts are the clustering result of the item-based concept space. The user-based concepts are the result of clustering adjustment from the item-based concepts with the information of users'' behaviors, such as whether or not a user is interested in both items in a concept. The user-base and item-based concepts co-evolve iteratively in the above manner. At the end, the proposed RS utilizes the learned concepts combined with the reading dependence to perform recommendation. The proposed techniques are demonstrated on the article recommendation. In this case, the features of an item correspond to the segmented contents of an article, and users'' behaviors correspond to users'' reading preferences. In the experiment, the item-based/user-based CF dimension is about $30,000$ and $3,000$ while the concept space in proposed RS articles starts from $5$ and ended up merely $87$ after $12$ iterations. The proposed RS dynamically adjust the dimension of articles. The dimensions of articles is $44$ in the end and used for clustering articles. New articles then can be clustered and recommended as well. The precision-recall curves indicates that the proposed RS achieves more hits than user-based/item-based CF and content-based filtering. The average precision-recall curves and mean average precision of proposed system grows and exceeds others. This idea of two concept spaces can be extended to the situation with items with extractable features as dimension and the interaction between items and users.