Research

Recommender systems

Recommender systems quickly and efficiently sort through vast quantities of information and bring the relevant pieces of information to our attention. They are quickly becoming essential to dealing with the information overload caused by the advent of the Internet. Many websites such as Amazon, NetFlix, and Last.FM generate recommendations for their users based on knowledge of the users’ interests (e.g. “Customers who bought this item also bought X”). So far, I’ve focused on the application and comparison of different recommendation algorithms in the news domain and in social bookmarking systems. I also wrote my Ph.D. thesis on this subject, along with several other publications:

  • Collaborative and Content-based Filtering for Item Recommendation on Social Bookmarking Websites. T. Bogers and A. Van den Bosch. In: Proceedings of the ACM RecSys ‘09 workshop on Recommender Systems and the Social Web, pages 9-16, 2009.
  • Recommending Scientific Articles using CiteULike. T. Bogers and A. van den Bosch. In RecSys ‘08: Proceedings of the 2008 ACM Conference on Recommender Systems, pp. 287-290. ACM Press, October 2008. [pdf]
  • Comparing and Evaluating Information Retrieval Algorithms for News Recommendation. T. Bogers and A. van den Bosch. To appear in Recommenders ‘07

Social bookmarking & folksonomies

One of the hip, new research areas that has emerged in the past five years is social tagging. Social tagging is the epitome of the internet as a democratic medium and is the process whereby resources are collectively labeled or categorized by the main users of a system (as opposed to the librarians of yore). The labels are commonly known as tags and the labeling process is called tagging. This collaborative tagging process is supposed to make a body of information increasingly easier to search, discover, and navigate over time. A collectively tagged body of resources is called a folksonomy (more or less). Some famous examples of websites using social tagging are Flickr, Delicious and CiteULike.

I’m interested in how we can use these collaboratively generated annotations to improve tasks like recommendation, search, and information seeking. I wrote my Ph.D. thesis on how recommender systems can be used to improve information access on social bookmarking websites. I’ve also looked at specific problems for social bookmarking systems such as spam detection and automatic detection of duplicate content. I’m currently in the middle of writing my dissertation on recommendation for social bookmarking. Some recent publications on this include:

  • Collaborative and Content-based Filtering for Item Recommendation on Social Bookmarking Websites. T. Bogers and A. Van den Bosch. In: Proceedings of the ACM RecSys ‘09 workshop on Recommender Systems and the Social Web, pages 9-16, 2009.
  • Using Language Modeling for Spam Detection in Social Reference Manager Websites T. Bogers and A. van den Bosch. In R. Aly, C. Hauff, I. den Hamer, D. Hiemstra, T. Huibers, and F. de Jong (Eds.), Proceedings of the 9th Belgian-Dutch Information Retrieval Workshop (DIR 2009), pp 87-94. Enschede, Februari 2009. [pdf]
  • Recommending Scientific Articles using CiteULike. T. Bogers and A. van den Bosch. In RecSys ‘08: Proceedings of the 2008 ACM Conference on Recommender Systems, pp. 287-290. ACM Press, October 2008. [pdf]
  • Using Language Models for Spam Detection in Social Bookmarking T. Bogers and A. van den Bosch. In Proceedings of 2008 ECML/PKDD Discovery Challenge Workshop, pp. 1-12, Antwerp, September 2008. [pdf]

Earlier in my PhD I also briefly investigated expertise tagging: the application of social tagging to the problem of finding experts. A related publication about this is:

  • Expertise Classification: Collaborative Classification vs. Automatic Extraction. T. Bogers, W. Thoonen, and A. van den Bosch. In J. Turner and J. Tennis (Eds.) Proceedings of the 17th ASIS&T SIG/CR workshop on Social Classification, 2006. [pdf]

In the context of my work on expert search I’ve also looked at Webwijs, our university’s repository of experts and expertise areas, developed for PR purposes. Out of curiosity I generated sometag clouds for the top 100 most popular tags used by people. Here’s the Dutch version and here’s the English version.

Expert search

My research on information retrieval has focused mostly on expert search: given an information collection, how can you automatically determine the top experts on a specific topic?

Much of my research in this area has focused on Webwijs, our university’s repository of experts and expertise areas, developed for PR purposes. Together with Krisztian Balog and Katja Hofmann from the University of Amsterdam I have been looking at different aspects of expert finding at a university such as Tilburg University. We proposed and constructed an expert search test collection called the UvT Expert Collection for our experiments as an alternative for (or rather a complement to) the W3C collecion used in most of the expert retrieval work done so far. See our SIGIR 2007 paper, our SIGIR 2008 workshop paper, and our JASIST 2010 article for more information about this.

  • Contextual Factors for Similar Expert Finding. K. Hofmann, K. Balog, T. Bogers, and M. de Rijke. Journal of the American Society for Information Science, 2010.
  • Integrating Contextual Factors into Topic-Centric Retrieval Methods for Finding Similar Experts. K. Hofmann, K. Balog, T. Bogers, and M. de Rijke. In Proceedings of the SIGIR 2008 Workshop on Future Challenges in Expert Retrieval, pp. 29-36. Singapore, July 2008 [pdf]
  • Broad Expertise Retrieval in Sparse Data Environments. K. Balog, T. Bogers, L. Azzopardi, M. de Rijke, and A. van den Bosch. In SIGIR ‘07: Proceedings of the 30th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval, pp. 551-558. ACM Press, July 2007. [pdf]

Together with a former Master’s student Ruud Liebregts, we also developed a univerity-wide expert search engine and subjected it to thorough system-based and user-based evaluation. You can try out the prototype version of the expert search engine here. See our ECIR 2009 paper for more information about this work, along with my other publications on this topic:

  • Design and Evaluation of a University-wide Expert Search Engine R. Liebregts and T. Bogers. In Proceedings of the 31st European Conference on Information Retrieval (ECIR 2009), vol. 5478 of Lecture Notes on Computer Science, pp. 587-594. Springer Verlag, April 2009. [pdf] [poster]
  • Using Citation Analysis for Expert Retrieval in Workgroups. T. Bogers, K. Kox, and A. van den Bosch. In E. Hoenkamp, M. de Cock, and V. Hoste (Eds.), Proceedings of the 8th Belgian-Dutch Information Retrieval Workshop (DIR 2008), pp 21-28. Maastricht, April 2008. [pdf]
  • Expertise Classification: Collaborative Classification vs. Automatic Extraction. T. Bogers, W. Thoonen, and A. van den Bosch. In J. Turner and J. Tennis (Eds.) Proceedings of the 17th ASIS&T SIG/CR workshop on Social Classification, 2006. [pdf]
  • Authoritative re-ranking of search results. Bogers, T. and Van den Bosch, A. In Proceedings of the 28th European Conference on Information Retrieval (ECIR 2006), vol. 3936 of Lecture Notes on Computer Science, pp. 519-522. Springer Verlag, April 2006. [pdf] [poster]
  • Authoritative re-ranking in fusing authorship-based subcollection search results. Bogers, T. and Van den Bosch, A. In F. de Jong and W. Kraaij (Eds.), Proceedings of the Sixth Belgian-Dutch Information Retrieval Workshop (DIR 2006), pp 49-55. Enschede: Neslia Paniculata, March 2006. [pdf]

Personal Information Management Agents

My PhD was part of the À Propos project, which focused on building a better information management agent. The goal of agent was to recommends interesting and relevant related literature based on the current writing task of the user. The initial motivation for my work on expert search stems from this project. The following publications shed some more light on the goals of the À Propos project:

  • A Personalized Recommender System for Writing in the Internet Age. M. C. Puerta Melguizo, O. Muñoz Ramos, L. Boves, T. Bogers, and A. Van den Bosch. In: Proceedings of the LREC 2008 workshop on Natural Language Processing Resources, Algorithms, and Tools for Authoring Aids, 2008.
  • What a Proactive Recommendation System Needs: Relevance, Non-Intrusiveness, and a New Long-Term Memory. M. C. Puerta Melguizo, T. Bogers, A. Deshpande, and A. Van den Bosch. In: Proceedings of ICEIS 2007, 2007.
  • À Propos: Pro-Active Personalization for Professional Document Writing. T. Bogers and A. van den Bosch. Long abstract in Proceedings of IIiX 2006, October 2006. [pdf] [abstract]