S-index implementation and performance for textbases

Abstract

In the present study we implement an improved, compressed variation of S-Index and consider its performance against a 130 MB textbase. S-Index, where ”S” stands for signature and ”Index” implies the inverted file index, is a hybrid indexing scheme found to combine advantages from two popular indexing methods: the inverted file and the signature file/bitmap. We present the file structure and consider implementation details relating to the improved compressed variation of S-Index. The results obtained are compared to those of the self-indexed compressed variation of the inverted file index. With minimal configuration/tuning effort, S-Index is measured to occupy less than 5% of the indexed textbase corpus. The latter implies performance comparable to that of file inversion.