A Method of User Modeling and Relevance Simulation in Document Retrieval Systems

摘要

Modeling users' information interests and needs is one of the most important task in the personalization of information retrieval domain. Search engines' algorithms are constantly improved but they still return a lot of irrelevant documents. For this reason a user is not able to quickly find a document with necessary information. In this paper we propose a user profile that is updated based on novel relevance judgment method. Proposed algorithms are a part of personalization agent system. As performed experimental evaluations have shown, the distance between current user profile and user preferences is decreasing when our algorithm is applied.