Research On Index Security And Cache Policy Of Distributed Search Engine

This work is supported by National High Technology Research and Development Program of China named"Key Research on Distributed Full Cipher Text Retrieval System", which is designed for government offices and large-scale enterprises with high demands of documents security. There are millions of documents and thousand of users need to maintain in these organizations, only distributed search engine can satisfy their needs. Meanwhile the system must give fully consideration to index security and access control to avoid the leakage of information.We have focused on these fields for a few years, released"Centralized/ Distributed Full Cipher Text Retrieval System"and successfully applied to e-government of Hubei Province. However, the former version had to face the threat of statistical attack and bear the slow response without cache. Our work was devoting to make improvements on these defects.On the one hand, we re-constructed the index to adapt the new security policy based on physical-block encryption. This policy ignores the logical units of index, encrypts the posting-list by the same physical length block, preventing malicious users from statistical attack follow the cipher block length difference.On the other hand, we added a distributed cache to searcher and indexer separately. Cache in searcher keeps the query results from former request, and is compatible with the access control index; cache in indexer holds the posting-lists of terms, splits into static part and dynamic part, the former one caches the high frequency words to satisfy the regular requests over the long term and the latter one caches the fresh words recently used to fulfill hot requests in the short term.