Thesis approved

I am happy to announce that my PhD thesis titled People Search in the Enterprise has been approved by the committee. The public PhD defense will take place on the 30th of September, 2008.
It is planned that the final version of thesis will be made available online early July, 2008.

Thesis completed

I am happy to announce that my thesis titled People Search in the Enterprise has been completed and submitted to the committee.

The main focus in the thesis is on two main expertise retrieval tasks: (1) expert finding — identifying a list of people who are knowledgeable about a given topic (“Who are the experts on topic X?”) and (2) expert profiling — returning a list of topics that a person is knowledgeable about (“What topics does person Y know about?”). In the thesis, expertise retrieval is approached as an association finding task between people and topics.

The main contribution of the thesis is a generative probabilistic modeling framework for capturing the expert finding and profiling tasks in a uniform way. On top of this general framework two main families of models are introduced, by adapting generative language modeling techniques for document retrieval in a transparent and theoretically sound way.

Throughout the thesis we extensively evaluate and compare these baseline models across different organizational settings, and perform an extensive and systematic exploration and analysis of the experimental results obtained. We show that our baseline models are robust yet deliver very competitive performance.

Through a series of examples we demonstrate that our generic models are able to incorporate and exploit special characteristics and features of test collections and/or the organizational settings that they represent. Additionally, we address a number of related tasks, including finding similar experts, mining contact details of people, and enterprise document search.

Finally, we provide further examples that illustrate the generic nature of our baseline models and apply them to find associations between topics and entities other than people.

Assuming that the committee’s answer is affirmative, the thesis is going to be printed in early June 2008.

Happy new year & welcome back

I took a little break from work so I could celebrate Christmas, spend time with the family, etc. I am back online now, and ready to commit myself to full-time thesis writing for the upcoming several weeks.

As to expert search material, here is a quick update.

  • Our (me and Maarten de Rijke) recent paper titled Associating People and Documents has been accepted to ECIR 2008. Common to most expertise search approaches is a component that estimates the strength of the association between a document and a people. In this paper we perform a careful analysis and investigation of how different association methods contribute to performance. The camera-ready version of the paper will be available from the Publications page, after jan 11).
  • We (me, Maarten, and Leif Azzopardi) submitted a paper titled A Language Modeling Framework for Expertise Search to the Information Processing and Management (IPM) journal. In this paper we introduce our language modeling approaches to expertise search in detail, and integrate these into a generative probabilistic framework. Since it is not a conference paper, it may take some time until it can be published.

There is some reading material from CIKM 2007:

Looks like the topic of expertise retrieval is gaining more and more popularity in IR conferences. While browsing the list of accepted papers for ECIR 2008, I found 3 full papers (out of 33) and 1 short paper (out of 19) about expert search, which gives the topic a solid presence.

  • (Serdyukov and Hiemstra)
    Modeling documents as mixtures of persons for expert finding [full]
  • (Balog and de Rijke)
    Associating People and Documents [full]
  • (Macdonald et al.)
    High Quality Expertise Evidence for Expert Search [full]
  • (Macdonald and Ounis)
    Expert Search Evaluation by Supporting Documents [short]

TREC Enterprise preparations

In the light of TREC conference preparations, I spent quite some time on performing additional experiments using the CSIRO collection this week. Although difficulties did occur, the road at the end has led to compelling results, with lessons learned. I will present our group’s results next Friday at TREC, and of course will put slides online after the talk.

While the problem of expertise management is not new in the Knowledge Management field, it was the introduction of the Expert search task at TREC 2005 that certainly attracted the attention of the IR community. This interest increased in 2006, and I anticipate it has just grown further this year. When I started my PhD career in Sept 2005, I immediately started working on expert finding, and looking back over the past two years, I feel this was a bet on the right horse.

Turning back to the TREC Enterprise track, on the other hand, the Discussion search task — featured in 2005 and 2006 — received much less interest. In fact, it was replaced with a Document search task this year. While it may be considered less innovative, I do find this interesting for the following reasons:

  • the (new) CSIRO collection is quite different from both the W3C and the WWW settings;
  • topics are broader than in case of W3C and topic definitions include example documents; and last but not least,
  • the same set of topics is used in document and expert search.

I am looking forward hearing about how others tackled this year’s Enterprise search tasks, what the organizers’ plans are for next year, and of course, curious to see how our approaches performed compared to that of other participants. While W3C became the de facto standard evaluation set for expert search, it only represents one type of intranet. Therefore, I hope that forthcoming publications on expert search will include results using CSIRO and the TREC 2007 platform too.