Research On Private Information Retrieval Algorithm

With the extensive application of information technology, publicly accessible databasesand search engineers are indispensable resources for retrieving the latest information.However, because of the shortcomings of traditional private information retrieval model, itis difficult to apply to large databases and search engines. Therefore, the research on newand practical private information retrieval model and algorithm has important significance.According to the analysis of the implemented private information retrieval systems andfunctional requirements of word semantic similarity-based private information retrievalsystem, we propose a new word semantic similarity-based private information retrievalmodel. Then we take related analysis to the parts of word semantic similarity computing,bogus keyword selection strategy, query information hiding, and the query results filteringand give the overall architecture of the system. The architecture of the system includesword semantic similarity computing modular, query processing modular and pagecrawling and filtering modular.We introduce the implement of word semantic similarity computing algorithm based onWordNet and HowNet. On the basis of the existing word semantic similarity computingalgorithm based on WordNet, we take the influencing factor of the depth of the node intoaccount. Then we apply this algorithm to the primitive similarity computing of HowNet.The experiments show that improved algorithm gets more accurate similarity results,according more with the human beings semantic habit.We introduce word semantic similarity-based private information retrieval algorithm.Bogus keyword selection criteria are the key point of the algorithm. The algorithmchooses word semantic similarity as the bogus keyword selection criteria, requiringsemantic similarity between bogus keyword with target keyword meets certain conditions.The time complexity of the algorithm is O (k), where k is the amount of boguskeywords. The experiments show that the query result quality of this model improves andthe information entropy decreases, In comparison with the GooPir model, but the declineof information entropy is within a small rate.