A novel semantic-aware search scheme based on BCI-tree index over encrypted cloud data

Most of the traditional privacy-preserving search schemes in cloud adopt TF-IDF model which is on the basis of keyword frequency statistics. The embedding semantic association between keywords and documents are not considered. To solve this problem, we propose a novel semantic-aware search scheme based on BCI-tree index over encrypted cloud data. The LDA model is adopted to generate vectors for documents and queried keywords and the vectors contain topic-based semantic information. The homomorphic encryption on vectors is used to perform privacy-preserving semantic relevance score computation between queried keywords and documents. To achieve efficient search processing, a novel binary clustering tree-based index (BCI-tree index) is designed, which is constructed following the divisive hierarchical clustering algorithm. By using the BCI-tree index, a depth-first recursive search algorithm is proposed. In addition, a threshold presetting-based optimization is applied to further accelerate the search speed. The experimental results show that the proposed scheme performs good in the semantic precision of search results and the search time cost.