Authors: Bai Long, Guang-Zhong Sun, Jing Yuan and Guoliang Chen
Real-time property and low power consumption has recently become two hot research topic for enhancing the performance of traditional search engines. This paper presents a real-time updating inverted index structure to improve the efficiency of search engines. This structure consists of main inverted index, auxiliary inverted index and content filtrating index, instead of the single model of traditional inverted index. Based on this inverted index structure, we implement a real BBS Search engine. According to the user feedbacks, this search engine has a significant improvement than the traditional method in terms of the real-time property of search results. To evaluate the efficiency and power consumption of our approach, we conduct extensive experiments on several high performance computing platforms. Experiments results show that the proposed inverted index structure has a 2 orders of magnitude improvement in terms of updating time than the traditional inverted index. Furthermore, our method saves 11% power consumption than the competing approach based on the Teraflops high performance green computing platform KD60.