Feedback model for microblog retrieval

Information searching in microblog services has become common and necessary for social networking. However, microblog retrieval is particularly challenging compared to web page retrieval because of serious vocabulary mismatch problem and non-uniform temporal distribution of relevant documents. In this paper, we propose a feedback model, which includes a feedback language model and a query expansion model considering both lexical expansions and temporal expansions. Experiments on TREC data sets have shown that our proposed model improves search effectiveness over standard baselines, lexical only expansion model and temporal only retrieval model. ? 2015, Springer International Publishing Switzerland, All rights Reserved.EI529-544904