Improving Precision in IR Considering Dynamic Environments

Abstract

Much of the research in Information Retrieval (IR) is devoted to studying the improvement of personalized results for specific users in a static environment. Nevertheless, few approaches take advantage of collective past searches in a dynamic context where the number of documents is increased according with the passage of time. In this paper, we present an on-line probabilistic algorithm, which uses the collective past searches in a dynamic context to answer static and dynamic queries. Several experiments were carried out with the aim of evaluating the effectiveness of our algorithm. The algorithm's results were compared with the cosine measure. Following the Cranfield paradigm, simulated datasets were used in the experiments. Final results show that it is possible to improve effectiveness in a dynamic context.