In-memory hash tables for accumulating text vocabularies

Abstract

The performance of several data structures for building libraries using a range of data collections and machines was investigated. The time and space required by three kinds of data structures namely, binary search trees (BST), splay trees, and hash tables was obtained. For splay trees, the time taken for a tree splayed at every access and every eleventh access was measured. In hash tables, the time taken with two hash functions was measured. The results show that move-to-front bit-wise hashing is the method of choice for efficient index construction.