This book deals with the improvement of user modeling in the context of Collaborative and Social Information Access and Retrieval (CSIRA) techniques. Information access and retrieval techniques are aimed at helping users to find information relevant to their needs. Today, in order to improve their effectiveness, some specific techniques have to take into account external characteristics such as those related to the user (and his context) which are most of the time little or not known by the system. Nevertheless we can observe that the applications related to the Web 2.0 which integrate users’ characteristics bring rather best results and at least personalized results.
It thus seems acquired that the collaborative and social aspects characterizing the users’ social context can be used to improve the way the information access and retrieval systems know each user through user modeling approaches.
Consequently improving user modeling taking into account social and collaborative aspects for information retrieval and access is a vast domain in which several disciplines intervene (computer science, cognitive science, information science, etc.). This is a recent research trend that integrates recommender systems, social networks analysis (Wasserman et al., 1994), adaptive information retrieval, user modeling, and social information retrieval (Kirsh, 2003) techniques and so on.
The objective of this book is to draw up a panorama of the concepts, techniques. and applications linked to CSIRA. This book is aimed at readers of any disciplines (information science or information technology, cognitive science, computer science, etc.) and contributes to the diffusion of the concepts to any public (graduate and post-graduate students, information system designers, information retrieval system designers, scientists, etc.).
Organization
This book presents operational and innovative ideas to integrate user modeling in order to improve CSIRA effectiveness. This book includes twelve chapters gathered in three sections. The first section covers generalities related to user modeling in the context. The second one deals with advances in user modeling. The third one presents some applications of such improved user modeling.
As it can be seen in the following description of the chapters, the contributions cover a large scope of techniques to improve user modeling in such a context. The reference section in each chapter includes numerous reference sources to help interested readers to find comprehensive sources and additional information.
Section I: User modeling in CSIRA
This section introduces one of the classical techniques of CSIRA and the way users are taken into account in these techniques. The two chapters are focused on recommender systems that are tools aiming at helping users to find items/information that they should consider as relevant from huge catalogues.
State-of-the-Art Recommender Systems describes the state of the art related to recommender systems. It deals with the three main types of filtering techniques and the way such techniques can be evaluated.
Computing Recommendations With Collaborative Filtering is focused on Collaborative Filtering techniques underlying different issues and system vulnerabilities. This chapter also presents a discussion related to when such techniques should be used, how recommendations are generated / evaluated.
Section II: Advances in User modeling in CSIRA
Through four chapters, this section is dedicated to the introduction of some advances in user modeling. Those advances are based on novel approaches taking into account communal tags, ontology-based semantic features, user intents and competencies.
Analyzing Communal Tag Relationships for Enhanced Navigation and User Modeling investigates methods for enabling improved navigation, user modeling and personalization using collaboratively generated tags. The authors discuss the advantages and limitations of tags, and describe how relationships between tags can be used to discover latent structures that can automatically organize a collection of tags owned by a community.
Adaptive User Profiles shows how a user profile can be built without any explicit input. Based on implicit behavior on social information networks, the profiles which are created are both adaptive (up to date) and socially connective. The proposed approach relies on the use of a Collaborative Tagging System like Delicious.
The authors of Modeling Users for Adaptive Information Retrieval by Capturing User Intent study and present their results on the problem of employing a cognitive user model for Information Retrieval (IR) in which a user’s intent is captured and used for improving his/her effectiveness in an information seeking task. The user intent is captured by analyzing the commonality of the retrieved relevant documents.
Ontology-Based User Competencies Modeling for E-Learning Recommender Systems explores a semantic Web-based modeling approach for document annotations and user competencies profile development. This approach is based on a same domain ontology set which constitutes the binder between materials and users. A variant of the nearest neighbor algorithm is applied to recommend concepts of interest and then document contents according to competencies profiles.
Section III: Improved user modeling – Application of CSIRA
This section supplies six chapters describing applications for which a specific user modeling is used to improve information retrieval and access techniques in a collaborative and social context. Such improved techniques are aimed at recommending more adapted bibliographical references, Web pages or music for instance and at adapting the information content to mobile users.
Combining information in synchronous collaborative IR environment explores the effectiveness of a sharing of knowledge policy on a collaborating group in order to satisfy a shared information need. The search engine exploits user relevance judgments to propose a new ranked list.
DemonD: a collaborative IR system build upon the actor network theory suggests a social search engine that identifies documents but more specifically users relevant to a query. It relies on a transparent profile construction based upon user activity, community participation, and shared documents.
COBRAS: Cooperative CBR Bibliographic Recommender System proposes a Peer-to-Peer bibliographical reference recommender system. It consists in finding relevant documents and interesting people related to the interests and preferences of a single person belonging to a like-minded.
Collaborative, content-based and case-based recommendation systems and their hybrids have been used for music recommendation. Music Recommendation by Modeling User's Preferred Perspectives of Content, Singer/Genre and Popularity shows how specific user information can be used to improve user model for ameliorate music recommendation accuracy.
Web content recommendation method based reinforcement learning is dedicated to a novel machine learning (based on reinforcement learning) perspective toward the web recommendation problem. A hybrid web recommendation method is proposed by making use of the conceptual relationships among web resources to derive a novel model of the problem, enriched with semantic knowledge about the usage behavior. The method is evaluated under different settings and it is shown how this method can improve the overall quality of recommendations.
Collaborating agents for adaptation to mobile users presents a twofold approach for adapting content information delivered to a group of mobile users. It is based on a filtering process which considers both the user’s current context and her/his preferences for this context.
Conclusion
The variety of the approaches developed to improve CSIRA effectiveness, as the richness of the various work undertaken on this subject tend to show that user modeling is in the center of the current concerns. In this way this book constitutes a real survey of advances and applications in user modeling for CSIRA.