Classification and Evaluation of Online Indexing Strategies

Most search engines have to face the dynamic nature of the web, and it becomes a big problem that how to offer near real-time query service while the underlying document collection increases dramatically every day. As a result, the online indexing approaches become one of the kernel research problems of information retrieval. In this paper, we first present a detailed classification of various online indexing strategies, from the classics to the state-of-the-arts. We then perform an evaluation on selected strategies. A new evaluation metric is introduced in this paper to characterize the dynamic performance when queries interact with online indexing concurrently. Evaluation results characterize the performance differences among strategies and indicate the future improvements on update and query performance.