Working with multiple representations of documents and information needs across different document types - recent work in integrated search and polyrepresentation
Decanato - Facoltà di scienze informatiche
Data d'inizio: 7 Giugno 2010
Data di fine: 8 Giugno 2010
The Faculty of Informatics is pleased to announce a seminar given by Christina Lioma and Birger Larsen
DATE: Monday, June 7th 2010
PLACE: USI Università della Svizzera italiana, room SI-008, Informatics building (Via G. Buffi 13)
As information becomes increasingly available in electronic form, we are faced with a demand from users for systems that seamlessly integrate content from many sources of different types, media and genres, and with different levels of available data. Such systems need to be capable of serving a plethora of different types of information needs. In particular, in the case of restricted domains, such as systems for retrieving highly technical or scientific information, it has been found that standard Web retrieval does not suffice, and that user information needs are notably different (and more demanding) than the average, short, keyword-based Web user information need. We present recent work that deals with several aspects of this challenge. The seminar will have two parts. In the first part, we will present work on developing a test collection for integrated search, consisting of different document types (domain-specific articles and books), with different levels of available data (fulltext, abstracts, and metadata) as well as several representations of user information needs (e.g. user keywords, user ideal answer, user's background) (Lykke et al., 2010). Ideas for a theoretical framework within which to integrate such representations as part of the principle of polyrepresentation (Ingwersen, 1996) will also be presented. In the second part of the seminar, we will present recent work in formalising the principle of polyrepresentation using subjective logic. A thorough experimental evaluation of this formalisation will also be presented.
Christina Lioma holds an M.Sc. in Natural Language Processing (with distinction) from the University of Manchester, and a Ph.D. in Computer Science (Information Retrieval) from the University of Glasgow. She has worked as a postdoctoral fellow at Katholieke Universiteit Leuven (Computer Science), and as a lecturer at Konstanz University (Linguistics). On April 2010 she joined Prof. Hinrich Schutze's NLP group at the University of Stuttgart. Research interests: Information Retrieval, Natural Language Processing, Computational Linguistics.
Birger Larsen is Associate Professor at the Royal School of Library and Information Science since 2006. From April 2010 he is leader of the research group on Information Systems and Interaction Design. His main research interests include Information Retrieval (IR), structured documents in IR, XML IR and user interaction, and exploitation of context in IR. From 2004 to 2007 he co-organized the Interactive Track at the Initiative for the Evaluation of XML Retrieval (INEX).
HOST: Prof. Fabio Crestani