A Research Of FTS Query Efficiency Based On Invert Index In RDBMS

These years, as a hotspot the research of full-text search get a lot of new achievements and breakthroughs. In particular, performance of text database has been greatly inproved, and it meets the most requirement of the Text Retrieval. However, it is too hard to meet requirement of the joint inquiries of full-text search and structure data retrieval for many text databases which has single function. On the other hand, except some expensive commercial RDBMS (Oracle, DB2, etc.), open source RDBMS represented by PostgreSQL make poor performance on full-text search. So that it has great significance to find a low-cost approache to meet the double requirements of full-text search and structure data retrieval.In order to meet the requirements above-mentioned, this paper has studied the research achievement of full-text search, and then analyzes PostgreSQL full-text search method which is based on inverted index. After this analysis, we found there is a great room of its inquery performance improvment. With above learning and research of index structure and retrieval processes of PostgreSQL full-text search, we found several reasons for its poor performance, and bring forward a program which is based on inverted index to improve the performace of PostgreSQL full-text search. This program includes the structure of inverted index transformation, scan method and so on.Finally, as showed in comparison test, the performance of PostgreSQL FTS has been greatly improved. The improved PostgreSQL FTS meets the double requirements of full-text search and retrieval of structure data. It's also proved the feasibility of insidely improving open source relational database to meet the double requirements.