Improving on S-Index: A Hybrid Indexing Scheme for Textbases

Abstract

We present two variations of S-Index and consider their storage utilization efficiency against a 130MB textbase. S-Index is a hybrid-indexing scheme that combines advantages from two popular indexing methods: the inverted file and the signature file. We introduce a new variation of the method and describe the file structure plus the implementation details for both the original as well as for the new variation of S-Index. The performance results obtained are used to compare the two variations. The improved variation of S-Index is measured to utilize less than 5% of the storage utilized by the corresponding indexed textbase corpus, under certain circumstances. The original S-Index involves slightly worse space requirements, but the performance of both variations is comparable to that of the compressed inverted file.