The Human-Centered Computing (HCC) group of DIKU joined with the Centre for Communication and Computing (CCC) are happy to announce a talk by visiting researcher Dr. Theodora Tsikrika, on search log data analysis by search engines. The talk is open to all.

Place: KUA, building 27, room 27.1.49 (Njalsgade 136, 2300 Copenhagen S)

Abstract

Search engines unobtrusively record in their logs large amounts of a broad range of user-system search interactions. Search logs have been extensively analysed so as to gain an understanding of users' searching behaviour, while the clickthrough data recorded in them have been exploited in several Information Retrieval applications by being interpreted as users' implicit relevance feedback. In the first part of this talk, I will present our research on search log analysis methods. Current methods describe the main features of the observed interactions in purely statistical terms without considering the semantics of the available information. We propose a semantic search log analysis method that enriches current approaches by exploiting the knowledge in a linked data cloud; particular focus has been placed on the analysis of users’ behavioural patterns regarding query formulation and modification. A study of the search logs of the commercial picture portal of a European news agency that were analysed using the proposed method and the implications of our findings will also be presented. In the second part of the talk, I will present our research on the exploitation of clickthrough data for image annotation and retrieval applications. We have explored (i) the use of clickthrough data for automatically generating labelled samples for training concept classifiers that perform image annotation, and (ii) the efficacy of topic modelling of clickthrough data in the image search domain.

Speaker's Profile:

Dr. Theodora Tsikrika received her PhD in Information Retrieval from Queen Mary University of London, under the supervision of Professor Lalmas. In 2007, she joined Centrum Wiskunde & Informatica in Amsterdam for a three-year period as a Researcher working on the VITALAS (video & image indexing and retrieval in the large scale) EU funded IP. In 2011, she joined the University of Applied Sciences Western Switzerland as a Researcher to work on the PROMISE (participative research laboratory for multimedia and multilingual information systems evaluation) and CHORUS+ (coordinated approach to the European effort on audio-visual search engines) EU funded projects. Her research interests are in the area of (multimedia) information retrieval, with particular focus on evaluation issues and on modelling the combination of evidence, Web information retrieval applications, and mining and analysis of search interactions logs. Since 2007, she has been involved in the coordination of international evaluation benchmarks (INEX, ImageCLEF).
Homepage: http://www.theodoratsikrika.info/