Combining Retrieval Results for Balanced Effectiveness and Efficiency in the Big Data Search Environment

In the big data age, we have to deal with tremendous amount of information, which is collected from various types of sources. For information retrieval systems, the collection of documents becomes larger and larger. For some query, an information retrieval system needs to retrieve a large number of documents as the result to the query. In reality, very often people mainly care about some top-ranked documents rather than the complete long list of documents. In such a situation, how to develop a retrieval system with desirable efficiency and effectiveness is a research problem. In this paper, we focus on the data fusion approach to information retrieval, in which each component retrieval system contributes a result and all the results are combined by a combination method. The goal of this research is to find a feasible combination method that is able to balance effectiveness and efficiency. Using 3 groups of historical runs from TREC for the experiment, we find that with the weights trained by weighted linear regression, the linear combination method can achieve good results in effectiveness and efficiency.