In this paper we address the problem of fuzzy search over encrypted data that supports misspelled search terms. We advance prior work by using a bit vector for bigrams directly instead of hashing bigrams into a Bloom filter. We show that we improve both index building performance as well as retrieval ratio of matching documents while providing the same security guarantees. We also compare fuzzy searchable encryption with exact searchable encryption both in terms of security and performance.
Göge, C., Waage, T., Homann, D., & Wiese, L. (2017). Improving fuzzy searchable encryption with direct bigram embedding. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 10442 LNCS, pp. 115–129). Springer Verlag. https://doi.org/10.1007/978-3-319-64483-7_8