An Efficient Method for Personalized Searching on Large Scale Data

In recent years, the information retrieval services on large-scale data in personal terminals have considerably increased. Many real end-users use a very small part in the large scale whole data for information retrieval. Furthermore, the reference frequencies of the used data are various according to end-user preference. However, the general indexes such as B+-tree and R-tree used in the services do not consider the searching patterns of individual users. Therefore, this paper proposes a new personalized index based on disk. The proposed index achieves much faster searching time than the existing indexes and also guarantees uniformed searching time over all data as a balanced tree. Our experiments show the effectiveness of our proposed method under various settings.