Using Query Expansion in Manifold Ranking for Query-Oriented Multi-Document Summarization

Using Query Expansion in Manifold Ranking for Query-Oriented Multi-Document Summarization Quanye Jia Beihang University jiaquanye@buaa.edu.cn Rui Liu Beihang University lr@buaa.edu.cn Jianying Lin Beihang University jianying.lin@buaa.edu.cn Abstract Manifold ranking has been successfully applied in query-oriented multi-document summariza- tion. It not only makes use of the relationships among the sentences, but also the relationships between the given query and the sentences. However, the information of original query is often insufficient. So we present a query expansion method, which is combined in the manifold rank- ing to resolve this problem. Our method not only utilizes the information of the query term itself and the knowledge base WordNet to expand it by synonyms, but also uses the information of the document set itself to expand the query in various ways (mean expansion, variance expansion and TextRank expansion). Compared with the previous query expansion methods, our method combines multiple query expansion methods to better represent query information, and at the same time, it makes a useful attempt on manifold ranking. In addition, we use the degree of word overlap and the proximity between words to calculate the similarity between sentences. We per- formed experiments on the datasets of DUC 2006 and DUC2007, and the evaluation results show that the proposed query expansion method can significantly improve the system performance and make our system comparable to the state-of-the-art systems. 1 Introduction Query-focused multi-document summarization is to create a summary from a set of documents that an- swers the information requirements expressed in the query. Compared to generic summarization, query- focused summarization requires the summary biased to a specific query besides the general requirement for a summary. In contrast to the task of question answering (QA) that mainly focuses on simple factoid questions and results in precise answers such as person, location or date, etc., in the case of query-focused summarization, the queries are mostly real-world complex questions and the information provided by the queries is insufficient. So how to understand the query and expand the query has a certain impact on the quality of the query-oriented summary. Currently most query oriented abstract systems are based on general summarization systems. By incorporating features that are related to the given query (e.g., the relevance of a sentence to the query), a generic summarization system can be adapted to a query-focused one. Due to the limit of information that a query can express, some systems also expand the query using some external resources such as WordNet, by which the synonyms of the query words can be obtained as expansion words (Zhou et al., 2005). However, such approaches to query expansion are restricted in query itself and external resources, because they cannot be applied to words not in WordNet such as named entities which frequently occur in queries, and much context related information cannot be captured by only synonyms. Lin(Lin et al., 2009) uses graph-based sentence ranking and sentence-to-word relationships to implement query expansion, which can improve the effect of graph ranking. However, the paper uses a single query expansion …

Content truncated to 500 words