Long and Difficult Query Processing Algorithm based on Dependency Relationship
Zhang Da-fu · Computer Knowledge and Technology · 2012
It is getting more and more attention to processing verbose query in search engine at present,but the performance of most com mercial and academic search engines deteriorates while handling longer queries.The problem that loss of focus on the main topics of the verbose query in the retrieval results is one of the main factors that hurt the average retrieval performance.This paper extend the traditional bag of idea.Every document is regarded as bag of which is composed of sentences.By dependency relation analysis,we get the dependence relationship of the words from the various sentences of bag of and long queries.And then we were ac cording to the matching degree of the dependence relationship to get the similarity scores between long queries and documents.Finally,we re-rank the initial results.Comparing with the traditional methods,this method fully make use of dependency relationship information of verbose queries and documents.Experiment on a standard TREC corpus shows that our approach can improve retrieval effectiveness for verbose query and the low recall rate.For the low recall rate,the MAP and P@N of our approach have a significantly improvement.