A query expansion based on sentence and vector integration strategy

This paper proposes a novel query expansion method to improve the average precision of the original query for information retrieval. The scheme uses a graph-based ranking algorithm to choose sentences in a manner which is different from existing sentence-based query expansion methods. At the same time a synthesis strategy for sentences is also built to construct new queries. The proposed solution is analyzed using DUC09 test collection data for update summarization task. Overall evaluation results show that the proposed method improves performance by yielding more relevant information with less noise.