Abstract: When editing a word processing document, we may search the Web by using a term in the document as an initial query and then modifying the query by adding keywords extracted from the text surrounding the search term. There are query expansion methods which use the text surrounding the search term in the initial result to weight candidate keywords in the source document to modify the query. However, this approach may lead to topic drift. To solve the problem, the initial results were filtered first and only the results containing similar contexts as that in the source document were selected to help choosing additional keywords. Experiments show that this method can get more appropriate additional keywords than other methods.

Key words: information retrieval, query expansion, context

摘要： 在查询扩展方法中，如果通过查询结果中关键词的上下文来计算候选关键词的权重，将权重大的词作为查询扩展词，其候选关键词来源于文档中关键词的上下文，这种方法存在主题漂移的问题。为了解决这个问题，提出一种将初始查询结果过滤，只选择与源文档语境相似的搜索结果，来帮助选择查询扩展词的方法。实验结果表明该方法能获得更合适的查询扩展词。

关键词: 信息检索, 查询扩展, 上下文