Personalized Question Answering

A common problem in Question Answering - and Information Retrieval in general - is information overload, i.e. an excessive amount of data from which to search for relevant information. This results in the risk of high recall but low precision of the information returned to the user. In turn, this affects the relevance of answers with respect to the users’ needs, as queries can be ambiguous and even answers extracted from documents with relevant content may be ill-received by users if they are too difficult (or simple) for them. We address the issue by integrating a User Modelling component to personalize the results of a Web-based opendomain Question Answering system based on the user’s reading level and interests.