Bringing Information Retrieval Back To Database Management Systems

Information retrieval emerged as independent research area from traditional database management system more than a decade ago. This was driven by the increasing functional requirements that modern full text search engines have to meet. Current database management systems (DBMS) are not capable of supporting such flexibility. However, with the increase of data to be indexed and retrieved and the increasing heavy workloads, modern search engines suffer from scalability, reliability, distribution and performance problems. The DBMS have a long tradition in coping with these challenges. Instead of reinventing the wheel, we propose using current DBMS as backend to existing full text search engines. This way, we bring back both worlds together. We present a new and simple way for integration and compare the performance of our system to the current implementations based on storing the full text index directly on the file system.