Project 7: Transparent User Profiling for Professional Search

Host Organisation: The Leiden Institute of Advanced Computer Science (LIACS), University of Leiden, the Netherlands

Objectives (1) to investigate the potential of user profiling in the context of professional search; (2) to investigate and develop methods for automatically deducing a professional user profile from their interactions with the search engine; (3) to experiment with explainable legal search, explaining the factors that contribute to the relevance ranking in the interface.

Expected results: An analysis will be made of the individual differences between all users of the legal search engine LegalIntelligence (WK) who gave their informed consent for using the log data. The analysis is based on interaction log data, showing the distribution of queries, topics, clicked documents, and behavioural characteristics (such as query length, session length). Methods for user profiling will be developed based on the analysis of individual differences in user behaviour (topics and interactions). The newly developed methods integrate models for user profiling based on keywords and documents, and models for user behaviour, based on search engine interactions. The quality and effectiveness of the user profiles will be evaluated with the use of search engine log data, interaction simulations, and end-users from a focus group of legal experts at Wolters Kluwer. A special focus is on explainable search, the goal of which is to make explicit in the result page which factors of documents constitute their relevance.

This project has received funding from the European Union's Horizon 2020 research and innovation programme under the Marie Skłodowska-Curie grant agreement No 860721