基于差分进化算法的语义查询扩展技术

Semantic retrieval extension technique based on differential evolution algorithm

目的提高扩展词与用户查询在语义层面上的关联,解决歧义扩展问题。方法基于差分进化算法的语义查询扩展技术先利用领域本体提供的领域背景知识来获取候选扩展词集,然后通过分析用户日志来获取用户检索偏好信息,最后利用差分进化算法确定同用户检索意图最相符的扩展词集。结果比起前沿的局部上下文分析方法,基于差分进化算法的语义查询扩展技术能够确定更高质量的扩展词集。结论利用用户日志和本体中概念间的语义关系作为背景数据来过滤无关的扩展词可以有效提高后续语义扩展过程的效率,差分进化算法能够有效排除同用户检索意图无关的词集并确定高质量的扩展词集。

Purposes—To improve the quality of retrieval expansion results with user＇s initial query and solve the ambiguity extension problem.Methods—Based on semantic retrieval extension technique of differential evolution algorithm（DE）,the domain background provided by the domain ontology is firstly used to obtain the candidate set of an extended word;and then the information of user＇s search preference is determined by analyzing the user log;finally,DE is used to determine the set of extended words which is closest to user＇s retrieval intention.Results—Compared to Local Context Analysis（LCA）,a state-of-the-art word extension technique,the semantic retrieval extension technique based on DE can determine high-quality set of extended words.Conclusion—The method of utilizing user log and the semantic relationship inside the ontology to filter the unrelated words can improve the efficiency of subsequent word extension process,and DE can effectively remove the words that falls outside of user＇s intension and determine high-quality set of extension words.