Contents -- Preface -- User Modeling and Profiling -- 1. Personalization-Privacy Tradeo s in Adaptive Information Access B. Smyth -- 1.1. Introduction -- 1.2. Case-Study 1 - Personalized Mobile Portals -- 1.2.1. The challenges of mobile information access -- 1.2.1.1. Mobile internet devices -- 1.2.1.2. Browsing versus search on the mobile internet -- 1.2.2. The click-distance problem -- 1.2.3. Personalized navigation -- 1.2.3.1. Pro ling the user -- 1.2.3.2. Personalizing the portal -- 1.2.4. Evaluation -- 1.2.4.1. Click-distance reduction -- 1.2.4.2. Navigation time versus content time -- 1.3. Case-Study 2: Personalized Web Search -- 1.3.1. The challenges of web search -- 1.3.2. Exploiting repetition and regularity in community- based web search -- 1.3.3. A case-based approach to personalizing web search -- 1.3.4. Evaluation -- 1.3.4.1. Successful sessions -- 1.3.4.2. Selection positions -- 1.4. Personalization-Privacy: Striking a Balance -- 1.5. Conclusions -- Acknowledgments -- References -- BIOGRAPHY -- 2. A Deep Evaluation of Two Cognitive User Models for Personalized Search F. Gasparetti and A. Micarelli -- 2.1. Introduction -- 2.2. Related Work -- 2.3. SAM-based User Modeling Approach -- 2.3.1. SAM: search of associative memory -- 2.3.2. The user modeling approach -- 2.3.2.1. LTS and STS -- 2.3.2.2. Sampling and Recovery -- 2.3.2.3. Learning -- 2.3.2.4. Interaction with Information Sources -- 2.3.3. HAL-based User Modeling Approach -- 2.4. Evaluation -- 2.4.1. Evaluating User Models in Browsing Activities -- 2.4.2. Corpus-based evaluation -- 2.4.3. Precision vs. Number of Topics -- 2.4.4. Precision vs. Extracted Cues -- 2.4.5. Precision vs. Size of STS -- 2.4.6. Precision vs. Number of Recovery Attempts -- 2.5. Conclusions -- References -- BIOGRAPHIES.

3. Unobtrusive User Modeling For Adaptive Hypermedia H. J. Holz, K. Hofmann and C. Reed -- 3.1. Introduction -- 3.1.1. User modeling in adaptive hypermedia -- 3.1.2. Motivation: informal education and the user modeling effect -- 3.1.3. Our solution: unobtrusive user modeling -- 3.2. Approach -- 3.2.1. Classi er-independent feature selection -- 3.2.2. Inference design -- 3.3. Field Study -- 3.3.1. ACUT -- 3.3.2. Measurements -- 3.3.3. Feature design -- 3.3.4. Data collection -- 3.3.5. Self-organizing maps -- 3.3.6. Revising the features -- 3.4. Discussion -- Acknowledgments -- References -- BIOGRAPHIES -- 4. User Modelling Sharing for Adaptive e-Learning and Intelligent Help K. Kabassi, M. Virvou and G. A. Tsihrintzis -- 4.1. Introduction -- 4.2. Description of Systems of Di erent Domains Sharing a Common User Model -- 4.2.1. System for e-Learning in Atheromatosis -- 4.2.2. Systems for Intelligent Help in le manipulation and e-mailing -- 4.2.3. Error Diagnosis in three systems of different domains -- 4.3. Common attributes for evaluating alternative actions -- 4.4. Example of a user interacting with three di erent sys- tems -- 4.5. User Modelling based on Web Services -- 4.5.1. UM-Server's Architecture -- 4.5.2. UM-Server's Operation -- 4.6. Multi-Attribute Decision Making on the Server side -- 4.7. Conclusions -- Appendix A. Multi-Attribute Decision Making -- References -- BIOGRAPHIES -- Collaborative Filtering -- 5. Experimental Analysis of Multiattribute Utility Collaborative Filtering on a Synthetic Data Set N. Manouselis and C. Costopoulou -- 5.1. Introduction -- 5.2. MAUT Collaborative Filtering -- 5.3. MAUT Algorithms for Collaborative Filtering -- 5.3.1. Proposed algorithms -- 5.3.1.1. Similarity per priority (PW) algorithm -- 5.3.1.2. Similarity per evaluation (PG) algorithm -- 5.3.1.3. Similarity per partial utility (PU) algorithm.

5.3.2. Nonpersonalized algorithms -- 5.4. Case Study and Experimental Analysis -- 5.4.1. Experimental setting -- 5.4.2. Results -- 5.5. Discussion -- 5.6. Conclusions -- References -- BIOGRAPHIES -- 6. Efficient Collaborative Filtering in Content-Addressable Spaces S. Berkovsky, Y. Eytani and L. Manevitz -- 6.1. Introduction -- 6.2. Collaborative Filtering -- 6.2.1. Reducing the computational effort required by the CF -- 6.3. Content-Addressable Data Management -- 6.4. CF over Content-Addressable Space -- 6.4.1. Mapping user pro les to content-addressable space -- 6.4.2. Heuristic nearest-neighbors search -- 6.4.3. Heuristic completions of user pro les -- 6.5. Experimental Results -- 6.5.1. Scalability of the search -- 6.5.2. Accuracy of the search -- 6.5.3. Inherent clustering -- 6.5.4. Completion heuristics -- 6.6. Conclusions and Future Research -- 6.6.1. Conclusions -- 6.6.2. Future research -- Acknowledgments -- References -- BIOGRAPHIES -- 7. Identifying and Analyzing User Model Information from Collaborative Filtering Datasets J. Gri th, C. O'Riordan and H. Sorensen -- 7.1. Introduction -- 7.2. Related Work -- 7.2.1. Weighting schemes in collaborative ltering -- 7.2.2. Graph-based approaches for recommendation -- 7.2.3. Collaborative ltering as a social network -- 7.3. Methodology -- 7.3.1. Collaborative ltering approach -- 7.3.2. Graph-based representations of the collaborative l- tering space -- 7.3.3. User features -- 7.4. Experiments -- 7.4.1. User model features -- 7.4.2. Spreading activation -- 7.5. Results -- 7.5.1. User model features -- 7.5.2. Spreading activation -- 7.6. Conclusions and Future Work -- References -- BIOGRAPHIES -- Content-based Systems, Hybrid Systems and Machine Learn- ing Methods -- 8. Personalization Strategies and Semantic Reasoning: Working in tandem in Advanced Recommender Systems Y. Blanco-Fern andez et al.

8.1. Introduction -- 8.2. Related Work -- 8.3. Our Reasoning Framework -- 8.3.1. The TV ontology -- 8.3.2. The User Pro les -- 8.3.2.1. Construction of the ontology-pro les -- 8.3.2.2. Level of interest of the users -- 8.3.3. A hybrid personalization technique -- 8.3.3.1. Content-based phase -- 8.3.3.2. Collaborative ltering phase -- 8.4. An Example -- 8.4.1. A hybrid recommendation by AVATAR -- 8.4.1.1. The content-based strategy in AVATAR -- 8.4.1.2. The collaborative strategy in AVATAR -- 8.5. Experimental Evaluation -- 8.5.1. Test algorithms -- 8.5.1.1. Approach based on association rules (Asso-Rules) -- 8.5.1.2. Item-based collaborative ltering approach (Item-CF) -- 8.5.1.3. Semantically enhanced item-based collaborative ltering (Sem-ItemCF) -- 8.5.2. Test data -- 8.5.3. Methodology and accuracy metrics -- 8.5.4. Assessment of experimental results -- 8.6. Final Discussion -- Acknowledgements -- References -- BIOGRAPHIES -- 9. Content Classification and Recommendation Techniques for Viewing Electronic Programming Guide on a Portable Device J. Zhu, M. Y. Ma, J. K. Guo and Z. Wang -- 9.1. Introduction -- 9.2. Related Work -- 9.3. Proposed System -- 9.3.1. Overview -- 9.3.2. EPG recommender system -- 9.4. Domain Identi cation and Content Recommendation -- 9.4.1. Classification problem and design choice -- 9.4.2. Maximum entropy model -- 9.4.3. Feature dimension reduction -- 9.4.4. Domain identification -- 9.4.5. Content classi er for recommendation -- 9.5. Prototype and Experiments -- 9.5.1. Prototype -- 9.5.2. Experimental database and protocol -- 9.5.3. Evaluation of ME classifier for domain identification -- 9.5.4. Evaluation of content recommendation -- 9.5.5. Preliminary evaluation of domain based recommen- dation -- 9.6. Conclusion -- References -- BIOGRAPHIES.

10. User Acceptance of Knowledge-based Recommenders Alexander Felfernig and Erich Teppan1 A. Felfernig, E. Teppan and B. Gula -- 10.1. Introduction -- 10.2. Koba4MS Environment -- 10.2.1. Architecture -- 10.2.2. Recommender knowledge base -- 10.2.3. Recommender process de nition -- 10.2.4. Calculating recommendations -- 10.3. Evaluation -- 10.3.1. Example application -- 10.3.2. Experiences from projects -- 10.3.3. Empirical ndings regarding user acceptance -- 10.4. Related Work -- 10.5. Summary and Future Work -- References -- BIOGRAPHIES -- 11. Using Restricted RandomWalks for Library Recommendations and Knowledge Space Exploration M. Franke and A. Geyer-Schulz -- 11.1. Motivation -- 11.2. Cluster Algorithms and Recommender Systems -- 11.3. Restricted Random Walk Clustering -- 11.3.1. Library usage data and similarity graphs -- 11.3.2. The restricted random walk stage -- 11.3.3. The cluster construction stage -- 11.3.4. Complexity -- 11.4. Giving Recommendations -- 11.5. Results -- 11.6. Updates of the Recommendation Lists -- 11.7. Conclusion -- Acknowledgments -- References -- BIOGRAPHIES -- 12. An Experimental Study of Feature Selection Methods for Text Classi cation G. Uchyigit and K. Clark -- 12.1. Introduction -- 12.2. Text Classi cation -- 12.3. Text Representation -- 12.4. Feature Selection -- 12.4.1. Filter Approach -- 12.4.2. Wrapper Approach -- 12.5. Feature Selection for Textual Domains -- 12.6. Feature Scoring Metrics -- 12.6.1. Chi-Squared statistic -- 12.6.2. NGL coefficient -- 12.6.3. GSS coefficient -- 12.6.4. Mutual information -- 12.6.5. Information gain -- 12.6.6. Odds ratio -- 12.6.7. Document frequency -- 12.6.8. Fisher criterion -- 12.6.9. BSS/WSS criterion -- 12.6.10. The GU metric -- 12.7. Experimental Setting -- 12.8. Empirical Validation -- 12.9. Results -- 12.10. Summary and Conclusions -- References -- BIOGRAPHIES.

Subject Index.