Relevance Feedback for Collaborative Retrieval Based on Semantic Annotations

A collaborative retrieval, based on the concept of sharing between users, is increasingly used to facilitate the research and to satisfy the needs. In this context, we suggest to improve the performance of collaborative research, taking account of the annotations as a new source of information describing the documents. In our contribution, we suggest to apply the relevance feedback to expand the user query. We also suggest a new approach based on co-occurrence to extract the relevant terms from annotations in the semi-structured documents returned by the collaborative retrieval systems. Experiments have shown their interest especially for the top ten documents returned by the system that have exceeded a value of 70% as the improvement rate. Keywords— Collaborative retrieval system, cooccurence, annotation, relevance feedback.