Information retrieval of mass encrypted data over multimedia networking with N-level vector model-based relevancy ranking

With an explosive growth in the deployment of networked applications over the Internet, searching the encrypted information that the user needs becomes increasingly important. However, the information search precision is quite low when using Vector space model for mass information retrieval, because long documents having poor similarity values are poorly represented in the vector space model and the order in which the terms appear in the document is lost in the vector space representation with intuitive weighting. To address the problems, this study proposed an N-level vector model (NVM)-based relevancy ranking scheme with an introduction of a new formula of the term weighting, taking into account the location of the feature term in the document to describe the content of the document properly, investigated into ways of ranking the encrypted documents using the proposed scheme, and conducted realistic simulation of information retrieval of mass encrypted data over multimedia networking. Results indicated that the timing of the index building, the most costing part of the relevancy ranking scheme, increased with the increase in both the document size and the multimedia content of the document being searched, which is in agreement with the expected. Performance evaluation demonstrated that our specially designed NVM-based encrypted information retrieval system is effective in ranking the encrypted documents transmitted over multimedia networks with large recall ratio and great retrieval precision.