Similar sentence extraction is an important issue because it is the basis of many applications. In this paper, we conduct comprehensive experiments on evaluating the performance of similar sentence extraction in a general framework. The effectiveness and the efficiency issues are explored on three real datasets, with different factors considered, i.e., size of data, top-
value. Moreover, the WordNet is taken into account as an additional semantic resource and incorporated into the framework. We thoroughly explore the performance of the updated framework to study the similar sentence extraction.