Unsupervised Query Segmentation Using Monolingual Word Alignment Method

In this paper, we propose a novel unsupervised approach to query segmentation using the word alignment model which is usually adopted in statistical machine translation system. Query segmentation is to obtain complete phrases or concepts in a query by segmenting a sequence of query terms, which is an important query processing procedure for improving information retrieval performance in search engines. In this work, we use a novel monolingual word alignment method to segment queries and automatically obtain the query structure in the form of multilevel segmentation. Our approach is language independent and unsupervised so that it is easy to be applied to various language scenarios. Experimental results on a real-world query dataset show that our approach outperforms the state of the art language model based method, which demonstrates the effectiveness of the proposed approach in query segmentation.