| Harnessing Models of Users' Goals to Mediate Clarification Dialog in Spoken Language Systems | | BIBAK | Full-Text | 3-13 | |
| Eric Horvitz; Tim Paek | |||
| Speaker-independent speech recognition systems are being used with
increasing frequency for command and control applications. To date, users of
such systems must contend with their fragility to subtle changes in language
usage and environmental acoustics. We describe work on coupling speech
recognition systems with temporal probabilistic user models that provide
inferences about the intentions associated with utterances. The methods can be
employed to enhance the robustness of speech recognition by endowing systems
with an ability to reason about the costs and benefits of action in a setting
and to make decisions about the best action to take given uncertainty about the
meaning behind acoustic signals. The methods have been implemented in the form
of a dialog clarification module that can be integrated with legacy spoken
language systems. We describe representation and inference procedures and
present details on the operation of an implemented spoken command and control
development environment called DeepListener. Keywords: Dialog systems; clarification dialog; spoken command and control; speech
recognition; conversational systems | |||
| Modeling the Acquisition of English: An Intelligent CALL Approach ☆ | | BIBA | Full-Text | 14-23 | |
| Lisa N. Michaud; Kathleen F. McCoy; Litza A. Stark | |||
| In this paper, we present a methodology for the development of a user model for CALL which captures various levels of language acquisition using individualized overlays supported with stereotypes. Our current focus is the empirical analysis of the order of written English grammatical structure acquisition in our learner population used to develop stereotype layers in our model. | |||
| Recognizing Time Pressure and Cognitive Load on the Basis of Speech: An Experimental Study | | BIBA | Full-Text | 24-33 | |
| Christian Müller; Barbara Großmann-Hutter; Anthony Jameson; Ralf Rummer; Frank Wittig | |||
| In an experimental environment, we simulated the situation of a user who gives speech input to a system while walking through an airport. The time pressure on the subjects and the requirement to navigate while speaking were manipulated orthogonally. Each of the 32 subjects generated 80 utterances, which were coded semi-automatically with respect to a wide range of features, such as filled pauses. The experiment yielded new results concerning the effects of time pressure and cognitive load on speech. To see whether a system can automatically identify these conditions on the basis of speech input, we had this task performed for each subject by a Bayesian network that had been learned on the basis of the experimental data for the other subjects. The results shed light on the conditions that determine the accuracy of such recognition. | |||
| The Learning Shell: Automated Macro Construction | | BIBAK | Full-Text | 34-43 | |
| Nico Jacobs; Hendrik Blockeel | |||
| By analysing sequences of actions performed by a user, one can find frequent
subsequences that can be suggested as macro (script) definitions. However,
often these 'actions' have additional features. In this paper we combine an
algorithm to detect frequent subsequences with an inductive logic programming
system to automatically generate for each frequent subsequence the most
specific 'template' for these additional features that is consistent with the
observed frequent subsequences. The resulting system is implemented and used in
an application where we automatically generate macros from logs of the use of a
Unix command shell. Keywords: machine learning; inductive logic programming; interface adaptation | |||
| Learning Interaction Models in a Digital Library Service | | BIBA | Full-Text | 44-53 | |
| Giovanni Semeraro; Stefano Ferilli; Nicola Fanizzi; Fabio Abbattista | |||
| We present the exploitation of an improved version of the Learning Server for modeling the user interaction in a digital library service architecture. This module is the basic component for providing the service with an added value such as an essential extensible form of interface adaptivity. Indeed, the system is equipped with a web-based visual environment, primarily intended to improve the user interaction by automating the assignment of a suitable interface depending on data relative to the previous experience with the system, coded in log files. The experiments performed show that accurate interaction models can be inferred automatically by using up-to-date learning algorithms. | |||
| A User Modeling Approach to Determining System Initiative in Mixed-Initiative AI Systems | | BIBAK | Full-Text | 54-63 | |
| Michael Fleming; Robin Cohen | |||
| In this paper, we address the problem of providing guidelines to designers
of mixed-initiative artificial intelligence systems, which specify when the
system should take the initiative to solicit further input from the user, in
order to carry out a problem solving task. We first present a utility-based
quantitative framework which is dependent on modeling: whether the user has the
knowledge the system is seeking, whether the user is willing to provide that
knowledge and whether the user would be capable of understanding the request
for information from the system. Examples from the application of sports
scheduling are included. We also discuss a qualitative version of the model,
for applications with sparse data. This paper demonstrates a novel use for user
models, one in which the system does not simply alter its generation based on
the user model, but in fact makes a user-specific decision about whether to
interact at all. Keywords: mixed-initiative systems; dialogue; exploiting user models to adapt
interaction; interactive scheduling; clarification; tailoring generation | |||
| Collaborating with Focused and Unfocused Users under Imperfect Communication | | BIBA | Full-Text | 64-73 | |
| Neal Lesh; Charles Rich; Candace L. Sidner | |||
| A totally focused user always finishes the current task or subtask before moving on to another. Typical users, however, sometimes shift back and forth between incomplete tasks and do not always communicate before doing so. This behavior poses a problem for a software agent that uses plan recognition to support its collaboration with users. Our solution is a discourse interpretation algorithm which balances between asking too many questions about a user's intentions and sometimes being wrong about them. | |||
| Improving User Modelling with Content-Based Techniques | | BIBAK | Full-Text | 74-83 | |
| Bernardo Magnini; Carlo Strapparava | |||
| SiteIF is a personal agent for a bilingual news web site that learns user's
interests from the requested pages.
In this paper we propose to use a content-based document representation as a starting point to build a model of the user's interests. Documents passed over are processed and relevant senses (disambiguated over WORDNET) are extracted and then combined to form a semantic network. A filtering procedure dynamically predicts new documents on the basis of the semantic network. There are two main advantages of a content-based approach: first, the model predictions, being based on senses rather then words, are more accurate; second, the model is language independent, allowing navigation in multilingual sites. We report the results of a comparative experiment that has been carried out to give a quantitative estimation of these improvements. Keywords: Content-Based User Modelling; Natural Language Processing; WORDNET | |||
| An Integrated Approach for Generating Arguments and Rebuttals and Understanding Rejoinders | | BIBAK | Full-Text | 84-94 | |
| Ingrid Zukerman | |||
| This paper describes an integrated approach for interpreting a user's
responses and generating replies in the framework of a WWW-based Bayesian
argumentation system. Our system consults a user model which represents a
user's beliefs, inferences and attentional focus, as well as the system's
certainty regarding the user's beliefs. The interpretation mechanism takes into
account these factors to infer the intended effect of the user's response on
the system's argument. The reply-generation mechanism focuses on the
identification of discrepancies between the beliefs in the user model and the
beliefs held by the system that are relevant to the inferred interpretation. Keywords: argumentation; Bayesian networks; plan recognition; discourse planning | |||
| Acquiring User Preferences for Product Customization | | BIBAK | Full-Text | 95-104 | |
| David N. Chin; Asanga Porage | |||
| Mass customization requires acquisition of customer preferences, which can
be modeled with multi-attribute utility theory (MAUT). Unfortunately current
methods of acquiring MAUT weights and utility functions require too many
queries. In Iona, the user is first queried for absolute/preferred constraints
and categorical preferences to cull the product pool. Next Iona selects queries
to maximally reduce the utility uncertainty of the remaining product choices.
Implemented queries include stereotype membership and contexts (the purchase
situation), which give probabilistic MAUT data modeled as ranges of weights.
The usefulness of a query is based on the reduction in uncertainty (smaller
range) weighted by the likelihood that the user belongs to a stereotype/context
based on similarity to the current user model. Querying proceeds until the
usefulness of the best query is below the threshold of user impatience. Finally
integer programming is used to select the best product for the user. Keywords: Multi-attribute utility theory; stereotype; user preferences; mass
customization; travel planning; user model acquisition; constraints | |||
| Utility-Based Decision Tree Optimization: A Framework for Adaptive Interviewing | | BIBA | Full-Text | 105-116 | |
| Markus Stolze; Michael Ströbel | |||
| An emerging practice in e-commerce systems is to conduct interviews with buyers in order to identify their needs. The goal of such an interview is to determine sets of items that match implicit requirements. Decision trees structure the interview process by defining which question follows a given answer. One problem related to decision trees is that changes in the selling strategy or product mix require complex tree restructuring efforts. In this paper we present a framework that represents the selling strategy as a set of parameters, reflecting the preferences of sellers and buyers. This representation of the strategy can be used to generate optimized decision trees in an iterative process, which exploits information about historical buyer behavior. Furthermore, the framework also supports advanced optimization strategies such as dynamic parameter adaptation and exit risk minimization. | |||
| User Modelling in I-Help: What, Why, When and How | | BIBAK | Full-Text | 117-126 | |
| Susan Bull; Jim Greer; Gordon McCalla; Lori Kettel; Jeff Bowes | |||
| This paper describes user modelling in I-Help, a system to facilitate
communication amongst learners. There are two I-Help components: Private and
Public Discussions. In the Private Discussions learners take part in a
one-on-one interaction with a partner (possibly a peer). The Public Discussions
are open -- everyone in the group has access to all discussion forums relevant
to that group. The Public Discussions are most suited to discussion of issues
where there might be a variety of valid viewpoints, or different solutions to a
problem. It is also useful for straightforward questions and answers that have
wide-spread applicability. The Private Discussions are better suited for more
intensive interactions involving peer tutoring or in-depth discussions. Because
there is only one helper in such situations, I-Help requires a method of
selecting an appropriate helper for an individual. We describe the user
modelling that takes place in each part of I-Help, in particular to effect this
matchmaking for Private Discussions. This modelling takes advantage of a
distributed multi-agent architecture, allowing currently relevant user model
fragments in various locations to be integrated and computed at the time they
are required. Keywords: peer help network; discussion forum; distributed student model | |||
| An Adaptive User-Interface-Agent Modeling Communication Availability | | BIBA | Full-Text | 127-136 | |
| Silke Höppner | |||
| Communication availability is closely connected to workspace awareness and
requires intelligent filtering, especially in Groupware-Systems. This paper
introduces a personalized adaptive communication availability agent, which is
currently in the process of implementation. The agent monitors user behavior,
predicts from there the current degree of user availability and then displays
it in a virtual world using an avatar. [1]
| |||
| Cognitive Computer Tutors: Solving the Two-Sigma Problem | | BIBA | Full-Text | 137-147 | |
| Albert Corbett | |||
| Individual human tutoring is the most effective and most expensive form of instruction. Students working with individual human tutors reach achievement levels as much as two standard deviations higher than students in conventional instruction (that is, 50% of tutored students score higher than 98% of the comparison group). Two early 20th-century innovations attempted to offer benefits of individualized instruction on a broader basis: (1) mechanized individualized feedback (via teaching machines and computers) and (2) mastery learning (individualized pacing of instruction). On average each of these innovations yields about a half standard deviation achievement effect. More recently, cognitive computer tutors have implemented these innovations in the context of a cognitive model of problem solving. This paper examines the achievement effect size of these two types of student-adapted instruction in a cognitive programming tutor. Results suggest that cognitive tutors have closed the gap with and arguably surpass human tutors. | |||
| Applying Interactive Open Learner Models to Learning Technical Terminology | | BIBAK | Full-Text | 148-157 | |
| Vania Dimitrova; John Self; Paul Brna | |||
| Our work explores an interactive open learner modelling (IOLM) approach
where learner diagnosis is considered as an interactive process involving both
a computer system and a learner that play symmetrical (to a certain extent)
roles and construct together the learner model. The paper presents an
application of IOLM for diagnosing and fostering a learner's conceptual
understanding in a terminological domain. Based on an experimental study, we
discuss computational and educational benefits of IOLM in terms of improving
the quality of the obtained learner model and fostering reflective thinking. Keywords: Intelligent tutoring systems; student modelling; meta-cognitive skills | |||
| Student and Instructor Models: Two Kinds of User Model and Their Interaction in an ITS Authoring Tool | | BIBA | Full-Text | 158-167 | |
| Maria Virvou; Maria Moundridou | |||
| WEAR is a Web-based authoring tool for Intelligent Tutoring Systems in Algebra related domains. Apart from modelling the student which is a common practice in almost all ITSs and ITS authoring tools, WEAR deals also with modelling the other class of its users: the instructors. Student and instructor models in WEAR interact with each other by exchanging information. This is in favour of both classes of WEAR's users, since they are affected by each other in a way similar to the one in a real educational setting. This paper describes the two kinds of user model and the type of information that they exchange. The issues raised in this research may be applied to other authoring t ools by the addition of an instructor modelling component. | |||
| METIORE: A Personalized Information Retrieval System | | BIBA | Full-Text | 168-177 | |
| David Bueno; Amos A. David | |||
| The idea of personalizing the interactions of a system is not new. With stereotypes the users are grouped into classes where all the users in a class have similar characteristics. Personalization was therefore not on individual basis but on a group of users. Personalized systems are also used in Intelligent Tutoring Systems (ITS) and in information filtering. In ITS, the pedagogical activities of a learner is personalized and in information filtering, the long-term stable information need of the user is used to filter incoming new information. We propose an explicit individual user model for representing the user's activities during information retrieval. One of the new ideas here is that personalization is really individualized and linked with the user's objective, that is his information need. Our proposals are implemented in the prototype METIORE for providing access to the publications in our laboratory. This prototype was experimented and we present in this paper the first results of our observation. | |||
| Personalizing Delivered Information in a Software Reuse Environment | | BIBAK | Full-Text | 178-187 | |
| Gerhard Fischer; Yunwen Ye | |||
| Browsing- and querying-oriented schemes have long served as the principal
techniques for software developers to locate software components from a
component repository for reuse. Unfortunately, the problem remains that
software developers simply will not actively search for components when they
are unaware that they need components or that relevant components even exist.
Thus, to assist software developers in making full use of large component
repositories, information access need to be complemented by information
delivery. Effective delivery of components calls for the personalization of the
components to the task being performed and the knowledge of the user performing
it. We have designed, implemented, and evaluated the CodeBroker system to
support personalized component delivery to increase the usefulness of a Java
software reuse environment. Keywords: Task modeling; discourse modeling; user modeling; software reuse;
information delivery | |||
| Automating Personal Categorization Using Artificial Neural Networks | | BIBA | Full-Text | 188-198 | |
| Dina Goren-Bar; Tsvi Kuflik; Dror Lev; Peretz Shoval | |||
| Organizations as well as personal users invest a great deal of time in assigning documents they read or write to categories. Automatic document classification that matches user subjective classification is widely used, but much challenging research still remain to be done. The self-organizing map (SOM) is an artificial neural network (ANN) that is mathematically characterized by transforming high-dimensional data into two-dimensional representation. This enables automatic clustering of the input, while preserving higher order topology. A closely related method is the Learning Vector Quantization (LVQ) algorithm, which uses supervised learning to maximize correct data classification. This study evaluates and compares the application of SOM and LVQ to automatic document classification, based on a subjectively predefined set of clusters in a specific domain. A set of documents from an organization, manually clustered by a domain expert, was used in the experiment. Results show that in spite of the subjective nature of human categorization, automatic document clustering methods match with considerable success subjective, personal clustering, the LVQ method being more advantageous. | |||
| User Modelling as an Aid for Human Web Assistants | | BIBA | Full-Text | 201-203 | |
| Johan Aberg; Nahid Shahmehri | |||
| This paper explores how user modelling can work as an aid for human assistants in a user support system for web sites. Information about the user can facilitate for the assistant the tailoring of the consultation to the individual needs of the user. Such information can be represented and structured in a user model made available for the assistant. A user modelling approach has been implemented and deployed in a real web environment as part of a user support system. Following the deployment we have analysed consultation dialogue logs and answers to a questionnaire for participating assistants. The initial results show that assistants consider user modelling to be helpful and that consultation dialogues can be an important source for user model data collection. | |||
| Modelling the Interests of a News Service User | | BIBA | Full-Text | 204-206 | |
| Fredrik Åhman; Annika Waern | |||
| We have developed a filtering service for an on-line news channel. In this domain, both content-based and collaborative filtering proved difficult to apply. Our solution was make extensive use of user involvement. In particular, we use the information gathered when users send tips about news articles to their friends. The paper describes the types of user involvement that our system allows, the techniques used for user modelling, and how these are used to generate relevant user-adaptive behaviour. | |||
| Category Based Customization Approach for Information Retrieval | | BIBAK | Full-Text | 207-209 | |
| Kenro Aihara; Atsuhiro Takasu | |||
| This paper proposes an customization technique to supporting interactive
document retrieval in unorganized open information space like WWW. We assume
that taxonomical thought is one of the most important and skilled operations
for us when we organize or store information. The proposed methodology,
therefore, handles hierarchical categories of documents. The system can be
customized through users' modification of categories. The features of the
proposed approach are (1) visualization of document categories for interaction,
(2) initialization of categories by hierarchical clustering method, (3)
customization of categories by support vector machine techniques, (4)
additional attributes for individual implicit cognitive aspects. Keywords: User Feedback; Customization; Visualization; Text Categorization;
Human-Computer Interaction; Support Vector Machine; Information Retrieval | |||
| Using Rollouts to Induce a Policy from a User Model | | BIBA | Full-Text | 210-212 | |
| Joseph E. Beck; Beverly Park Woolf | |||
| This research describes the application of an executable user model to generate policies to adapt software to best fit the user. Our approach first gathers data describing how users behave, and uses these data to induce a computational model that predicts how users will perform in a particular situation. Since system designers have differing goals, our architecture takes an arbitrary goal that the designer would like to see users achieve. Our architecture than using rollout techniques to determine how software should act in a particular situation with the user in order to achieve the desired goal. | |||
| Tailoring the Content of Dynamically Generated Explanations | | BIBA | Full-Text | 213-215 | |
| Kalina Bontcheva | |||
| This paper describes briefly an approach for tailoring the content of automatically generated hypertext explanations. The implemented pilot system hylite+ has a dynamically updated user model, which is used by the language generation modules to adapt the explanations to the user beliefs and preferences. The goal is to provide sufficiently detailed information which, on one hand helps the user by explaining the unknown terms, and on the other, avoids repeating already known facts. | |||
| A User Model Based on Content Analysis for the Intelligent Personalization of a News Service | | BIBAK | Full-Text | 216-218 | |
| Manuel de Buenaga Rodríguez; Manuel J. Maña López; Alberto Díaz Esteban | |||
| In this paper we present a methodology designed to improve the intelligent
personalization of news services. Our methodology integrates textual content
analysis tasks to achieve an elaborate user model, which represents separately
short-term needs and long-term multi-topic interests. The characterization of
user's interests includes his preferences about content, using a wide coverage
and non-specific-domain classification of topics, and structure (newspaper
sections). The application of implicit feedback allows a proper and dynamic
personalization. Keywords: Information dissemination; short/long-term models; multi-topic user profile;
adaptive user model; personalized information service | |||
| Modeling Exploratory Behaviour | | BIBAK | Full-Text | 219-221 | |
| Andrea Bunt; Cristina Conati | |||
| In this paper we propose a user model that aims to assess the user's
exploratory behaviour in an open environment. The model is based on a Bayesian
Network and consists of several components that allow diagnosis of the causes
of poor exploratory behaviour. Among these components are the user's knowledge
of exploration strategies, the user's motivation level, personality traits and
emotional states. Keywords: Bayesian Networks; exploration; exploratory environments | |||
| Ascribing and Weighting Beliefs in Deceptive Information Exchanges | | BIBAK | Full-Text | 222-224 | |
| Valeria Carofiglio; Fiorella de Rosis; Cristiano Castelfranchi | |||
| Humans apply a large variety of deception forms in their communicative acts;
they are not necessarily 'uncooperative' in doing so, as they may even deceive
for the benefit of their interlocutors [2,3]. Deception may be active or
passive, according to whether the Speaker does something or not, to achieve his
goal. It may be applied directly to the deception object p or may indirectly
influence it through some 'deception medium' q that may be a cause, an effect
or a diverting cause or effect of p. In this short paper, we examine how
deception may be simulated if mental states are represented as belief networks
and various weights are attached to beliefs. Keywords: belief ascription; belief networks; dialog simulation; deception | |||
| Visual Search and Background Complexity: Does the Forest Hide the Trees? | | BIBAK | Full-Text | 225-227 | |
| Martha E. Crosby; Marie K. Iding; David N. Chin | |||
| This research addresses the issue of cognitive complexity or cognitive load
in a visual search task. Eye tracking methodology was employed to track users'
eye fixations and scan patterns while counting targets in a visual array.
Background complexity and number of targets were varied. Results showed that
there was a positive relationship between fixation duration and background
complexity and between fixation duration and number of targets in the array.
Fixation duration and saccade predicted background complexity and number of
targets for simple and systematically varied arrays. These results indicate
that eye-tracking data may contribute effectively to the development of user
models in crisis management systems. Keywords: Eye-tracking; Cognitive-complexity | |||
| Characterizing Sequences of User Actions for Access Logs Analysis | | BIBA | Full-Text | 228-230 | |
| Thomas Draier; Patrick Gallinari | |||
| The paper presents new measures for characterizing sequences of user actions. They are aimed at categorizing user behavior on intranet sites. Their relevance is evaluated using different encoding and clustering algorithms. New criteria are introduced for comparing clustering methods. | |||
| Modeling Literary Style for Semi-automatic Generation of Poetry | | BIBAK | Full-Text | 231-233 | |
| Pablo Gervás | |||
| The generation of formal poetry involves both complex creativity -- usually
exercised by a human poet -- and strict algorithmic restrictions regarding the
metrical structure of the poem -- determined by literary tradition. Starting
from a generating system that enforces automatically the metrical restrictions,
this paper presents a model for the literary style of a user based on four key
features for user preferences -- word selection, language structures, poem
planning, and restrictions on realisation -- governing the generation of poetry
from input data provided by the user -- a prose paraphrase of the intended
message, a task specific vocabulary, and a corpus of construction patterns. The
system exploits the CBR paradigm as a means to evolve a case base (a vocabulary
/ construction pattern grouping) that effectively models the style of a
specific user as a result of multiple iterations through the CBR cycle. Keywords: natural language generation; human-computer collaboration; task modeling | |||
| Perceptual Considerations for Quality of Service Management: An Integrated Architecture | | BIBA | Full-Text | 234-236 | |
| George Ghinea; George D. Magoulas | |||
| In this paper, we suggest an integrated architecture that makes use of the objective-technical information provided by the designer and the subjective-perceptual information supplied by the user for intelligent decision making in the construction of communication protocols. Thus, this approach, based on the Analytic Hierarchy Process, incorporates not only classical Quality of Service (QoS) considerations, but, indeed, user preferences as well. Furthermore, in keeping with the task-dependent nature consistently identified in multimedia scenarios, the suggested communication protocols also take into account the type of multimedia application, which they are transporting. Lastly, our approach also opens the possibility for such protocols to dynamically adapt based on a changing operating environment. | |||
| Emotions and Personality in Agent Design and Modeling | | BIBA | Full-Text | 237-239 | |
| Piotr J. Gmytrasiewicz; Christine L. Lisetti | |||
| Our research combines the principled paradigm of rational agent design based on decision theory with formal definitions of the emotional states and personality of an artificial intelligent agent. We view the emotional states as the agent's decision-making modes, predisposing the agent to make its choices in a specific, yet rational, way. Change of the emotional state, say due to an external stimulus, invokes a transformation of the agent's decision-making behavior. We define personality as consisting of the agent's emotional states together with the specification of transitions taking place among the states. To model the personalities and emotional states of other agents and humans, we additionally provide a definition of a personality models of other agents. Our definition allows the personality models to be learned over the course of multiple interactions with the users and other agents. | |||
| Using Document Structures for Personal Ontologies and User Modeling* | | BIBAK | Full-Text | 240-242 | |
| Sanghee Kim; Wendy Hall; Andy Keane | |||
| We present a new approach that makes use of the embedded structural
information of the documents which a user frequently refers to for deriving a
personalized concept hierarchy and for identifying user preferences concerning
document searching and browsing. Keywords: Personal ontology; Supported browsing; Structured document | |||
| User-Tailored Plan Presentation | | BIBA | Full-Text | 243-246 | |
| Detlef Küpper; Alfred Kobsa | |||
| This paper discusses plan presentation, the second phase in user-tailored advice giving. Its main task is to determine what knowledge must be provided to ascertain that the user comprehends the plan and is able to perform it, even if he detects unexpected obstacles. Plan presentation is guided by a model of the user's knowledge and of his capabilities to perform actions in the domain. Finally we describe how to bias the plan generation process to prefer plans that contain as little information unfamiliar to the user as possible. | |||
| Investigating Students' Self-Assessment Skills | | BIBA | Full-Text | 247-250 | |
| Antonija Mitrovic | |||
| Student modeling approaches predominantly focus on modeling student knowledge. For effective learning, however, it is necessary to teach students how to learn, as well as to provide support for learning domain knowledge. Recently, a number of projects focused on students' learning strategies, and initiated work on modeling students' metacognitive skills, such as self-explanation and reflection. This paper focuses on self-assessment as an important metacognitive skill. We present results of an initial study carried out in the context of SQL-Tutor, a system that helps students to learn a database language. We found that not all students are good at evaluating their own knowledge, and that their knowledge level is an important factor. The study is an initial step towards incorporating the meta-level into the existing student model in SQL-Tutor. | |||
| Generating Personal Travel Guides -- And Who Wants Them? | | BIBA | Full-Text | 251-253 | |
| Cécile Paris; Stephen Wan; Ross Wilkinson; Mingfang Wu | |||
| In this paper we describe a system that generates synthesized web pages as a travel guide through integrating a discourse planner with a document retrieval system. We then present our investigation on whether the guide generated by such a system is actually preferred by users over a more general guide. | |||
| Inspectability and User Controlled Revision on Long Term User Models | | BIBA | Full-Text | 254-256 | |
| António Silva; Zita A. Vale; Carlos Ramos | |||
| Typically, the user models used in Intelligent Tutors tend to be tightly controlled by the system, due to the constraints imposed by the specific nature of the tutoring process. Therefore, no control by the user himself is usually allowed over his/her model's contents. In order to make the evaluation of the tutoring process a cooperative task adequate techniques should be devised. This paper describes early attempts to build a user model module for an Intelligent Tutor to be used in the training of electrical network Control Center operators. It also attempts to address the different requirements of the distinct phases of this tutoring environment. | |||
| Getting the Right Information to the Right Person | | BIBA | Full-Text | 257-259 | |
| Gloria Teo | |||
| Classical personalization methods require prior knowledge of user in order to adapt to user's needs. This paper presents an algorithm that generates an initial personal profile for new user with no prior knowledge of user's interest. | |||
| Goals, Tasks and Application Domains as the Guidelines for Defining a Framework for User Modelling | | BIBAK | Full-Text | 260-262 | |
| Ilaria Torre | |||
| The exponential increase of adaptive systems and the difficulty of
identifying the relevant dimensions for the construction of user models, are
the bases of the choice of building a framework for user modelling. The goal is
that of defining a method for identifying such dimensions and moreover creating
a taxonomy of goals, features and dimension for different application domains. Keywords: Framework and dimensions for user modelling; ontology for adaptation
features; reusable user modelling components; adaptive hypermedia | |||
| Improving Student Models by Reasoning about Cognitive Ability, Emotions and Gender | | BIBA | Full-Text | 265-267 | |
| Ivon Arroyo; Beverly Park Woolf | |||
| We explore how cognitive, socio-biological and emotional conditions of the student help predict behavior within an ITS, and how instruction should be adapted depending on these variables to improve educational outcomes. Cognitive, social and emotional factors tend to be more permanent in nature than student's knowledge. Our approach is to diagnose them with pre-tests before the user starts using the system. | |||
| Integrating Multilingual Text Classification Tasks and User Modeling in Personalized Newspaper Services | | BIBAK | Full-Text | 268-270 | |
| Alberto Díaz Esteban | |||
| In this paper a methodology designed to improve the intelligent
personalization of newspaper services is presented. The methodology integrates
textual content analysis tasks to achieve an elaborate user model, which
represents separately short-term needs and long-term multi-topic interests. The
characterization of user's interests includes his preferences about structure,
content and information delivery. A wide coverage and non-specific-domain
classification of topics and a personal set of keywords allow the user to
define his preferences about content. The application of implicit feedback
allows a proper and dynamic personalization. Another topic that have been
addressed in the thesis is the evaluation of systems offering to send users a
selection of the daily news by electronic mail. Finally, the extensions to a
multilingual framework are studied. Keywords: Short/long-term models; multi-topic user profile; adaptive user model;
evaluation; multilingual text classification tasks | |||
| Enhancing Embodied Intelligent Agents with Affective User Modelling | | BIBA | Full-Text | 271-273 | |
| Patrick Gebhard | |||
| The objective of this research is the exploration how affective knowledge used in global controlling mechanisms for public information systems with lifelike presentation agents will increase the effectiveness of such systems in terms of information presentation, but also help the user to better explain his/her needs by adopting a more natural conversational style through interactive dialogue. | |||
| Designing TV Viewer Stereotypes for an Electronic Program Guide | | BIBA | Full-Text | 274-276 | |
| Cristina Gena | |||
| This paper describes how a user modeling knowledge base for personalized TV servers can be generated starting from an analysis of lifestyles surveys. The aim of the research is the construction of well-designed stereotypes for generating adaptive electronic program guides (EPGs) which filter the information about TV events depending on the user's interests. | |||
| AGENDA CULTUREL An Adaptive Cultural Information Service | | BIBA | Full-Text | 277-279 | |
| Jessica Givone | |||
| In the context of Cartable Electronic project we developed a cultural information diffusion service. We describe the knowledge to define and exploit the user model. Then we propose a multi-agent architecture and we describe the system implementation. | |||
| Ubiquitous User Modeling for Situated Interaction | | BIBA | Full-Text | 280-282 | |
| Dominik Heckmann | |||
| The main contribution of my doctoral proposal will be the design of a standardized and expandable XML-based User Modeling Ontology Language, which enables ubiquitous systems to communicate about user models. The second contribution will be the investigation of combining simple partial user models from the point of view of the user modeling ontology language, as well as the specific example domain of speech and manual input, which will be realized by object-oriented dynamic Bayesian networks. | |||
| An Intelligent Pedagogical Agent in CALL | | BIBA | Full-Text | 283-285 | |
| Ognian Kalaydjiev | |||
| The contribution presents a planning agent in an adaptive Web-environment for second language terminology learning. This pedagogical agent provides active or passive sequencing and suggests system-learner dialogue by open learner model. It performs as well task-specific personalized information retrieval of relevant readings. Thesis synopsis. Thesis topic is "Intelligent Agents in Natural Language Processing Applications". The idea is to design and implement agents in information extraction, retrieval and filtering, and to apply them in a number of projects dealing with language technologies. | |||
| How to Learn More about Users from Implicit Observations1 | | BIBA | Full-Text | 286-288 | |
| Ingo Schwab | |||
| In this paper, an approach to learning user interests is presented. It relies on positive evidences only, in consideration of the fact that users rarely supply the ratings needed by traditional learning algorithms, specifically not negative examples. Learning results are explicitly represented to account for the fact that in the area of user modeling explicit representations are known to be considerably more useful than purely implicit representations. A content-based recommendation approach is presented. The described framework has been extensively tested in an information system. | |||
| Student Modelling for CALL Based on Pedagogical Standards | | BIBA | Full-Text | 289-291 | |
| Monika Tarantowicz-Gasiewicz | |||
| The doctoral dissertation summarised in this paper contains a proposal of pedagogical standards for the design and evaluation of a student model in a CALL system, and of such a system itself. The standards were derived from the Theory of Versatile Education. Upon these standards, a theoretical design of a student model in a CALL system was proposed. It was demonstrated how the potential of educational tradition and SLA research can be applied to CALL software for the sake of its pedagogical quality. | |||
| Evaluation of Adaptive Systems | | BIBA | Full-Text | 292-294 | |
| Stephan Weibelzahl | |||
| Unambiguously, adaptive systems have to be evaluated empirically to
guarantee that the adaptivity really works. Nevertheless, only few of the
existing adaptive systems have been evaluated. One of the most important
reasons for this lack is, that measures for adaptivity success have not been
investigated systematically up to now.
The aim of this PhD thesis is to explore a methodology for the empirical evaluation of adaptive systems, including validated criteria, experimental designs and procedures. It will be demonstrated that cognitive and behavioral factors provide important evidence for adaptivity success. | |||
| Supporting Negotiated Assessment Using Open Student Models | | BIBAK | Full-Text | 295-297 | |
| Juan-Diego Zapata-Rivera | |||
| During the last two years our research on open student models have led us to
experiment with visualization and inspection of Bayesian student models --
ViSMod, and employing conceptual maps as a representation of the student
through ConceptLab, a knowledge construction and navigation system. Although
previous work have given us interesting results, several questions remain to be
solved, such as: How open should be the student model to better enhance the
learning process?; How should students and teachers interact with the model?;
Are students and teachers willing to interact with the model or it should be
done as part of a supervised learning activity? In order to solve some of these
questions, this thesis explores different scenarios in which open student
modelling can be used as means of supporting reflection, negotiated assessment,
and enhance the learning process. Keywords: Open Student Modelling; Visualizing and Inspecting Bayesian Student Models;
Conceptual Maps; and Negotiated Assessment | |||
| Using Markov Chains for Structural Link Prediction in Adaptive Web Sites | | BIBAK | Full-Text | 298-300 | |
| Jianhan Zhu | |||
| My research investigates into using Markov chains to make link prediction
and the transition matrix derived from Markov chains to acquire structural
knowledge about Web sites. The structural knowledge is acquired in the form of
three types of clusters: hierarchical clusters, reference clusters, and grid
clusters. The predicted Web pages and acquired Web structures are further
integrated to assist Web users in their navigation in the Web site. Keywords: Markov chains; grid clusters; hierarchical clusters; reference clusters | |||
| Tailoring Privacy to Users' Needs 1 | | BIBA | Full-Text | 301-313 | |
| Alfred Kobsa | |||
| This article discusses how the deployment of personalized systems is affected by users' privacy concerns and by privacy legislation. It shows that these impacts are substantial and will require a significant enhancement of current systems. Basic requirements can already be met with existing technology. Most privacy laws however also impose demands that call for new technologies that still need to be researched. A central conclusion of the paper is that a uniform solution for privacy demands does not exist since both user preferences and legal stipulations are too heterogeneous. Instead, privacy will have to be dynamically tailored to each individual user's needs, and to the jurisdiction at both the location of the personalized system and that of the user. | |||
| Heavyweight Applications of Lightweight User Models: A Look at Collaborative Filtering, Recommender Systems, and Real-Time Personalization | | BIBA | Full-Text | 314 | |
| Joseph A. Konstan | |||
| Real-time personalization is one of the goals behind building effective and
accurate user models. A wide variety of applications, from graphical user
interfaces to information filtering and retrieval systems to electronic
commerce displays, can better serve users if they adapt to user wants and
needs. Many personalization systems take a heavyweight approach to
personalization-extensive modeling of the problem domain, user tasks, and user
preferences. While these heavyweight models can be very successful, they can be
extremely challenging to build and adapt, and the effort involved can lead
system designers to "cut corners" and reduce their fidelity and effectiveness.
A different approach, and one that is used in a wide variety of commercial recommender systems, is to build lightweight models that are almost entirely generic, and to rely on the presence of large amounts of data to overcome the reduced "intelligence" of the model. Collaborative filtering, the technology behind many of these recommenders, is often implemented using only minimal assumptions (i.e., that user preferences are relatively consistent) and highly generic data (i.e., preference or purchase data). This talk explores recommender systems, focusing on their strengths and limitations when used to achieve real-time personalization, and comparing them with richer, knowledge-based models. This exploration includes a tour of deployed research and commercial applications, viewed from both the developer's and the user's perspective. | |||
| Eye Tracking: A Rich Source of Information for User Modeling | | BIBA | Full-Text | 315 | |
| Sandra P. Marshall | |||
| This presentation focuses on two aspects of eye-tracking research: using
point-of-gaze information to look at shifts in attention and using changes in
pupil diameter to know when cognitive demands occur. Both aspects are
immediately important in cognitive modeling because as modelers, we want to
understand as much as we can about individuals' underlying cognitive processes.
Eye tracking yields very precise behavioral and physiological data that
contribute to our understanding.
The presentation describes several research examples in which eye-tracking data have been incorporated into basic cognitive models. Video clips illustrate the basic tasks and the ways that individuals' eyes respond when they attempt to complete the tasks. Graphic representations highlight and quantify the eye movements, providing colorful traces that show the position of the point-of-gaze every 4 msec. And, plots of pupil change over time show the correlation between effortful cognitive processing and stimulus difficulty. Research examples range from simple arithmetic calculations to complex military situations. The emphasis is on the ways that eye-tracking information can facilitate the modeling process, and the strengths and weakness of eye tracking methodologies are examined. Finally, the presentation showcases a few commercial applications in which eyetracking analyses have yielded important insights into human behavior. These examples show a vital connection between theoretical development of cognitive models and practical adoption of their results. | |||