High Precision Information Retrieval with Multi-Term Relation

Enterprise search differs from Internet search in many ways. The notion of a “good” answer to a query is quite different. On an intranet, the notion of a “good” answer is often defi ned as the “right” answer. Finding the right answer is often more diffi cult than fi nding the best answer. On the other hand, nowadays when searching on Internet, a large amount of information returned for a query, the most important is how to improve the precision of a query. In order to improve the search relevance and precision, a search approach integrated multi-term relations into search process is presented in this paper. In this approach, two level mixed indices are established including term index and term relations index. The Maximum Match algorithm in Chinese segmentation is used to fi nd multi-term relations from text data. With this approach, we can obtain the “right” answer quickly and effi ciently.