Using inverted files to compress text

Abstract

This is the first report on a new approach to text compression. It consists of representing the text file with compressed inverted file index in conjunction with very compact lexicon, where lexicon includes every word in the text. The index is compressed using standard index compression techniques, and lexicon is compressed with original dictionary compression method that gives better compression results than existing procedures. Compression procedure is complex, but decompression time is linear with the file size, although it requires two passes and hence can not be performed online. First experiments show that this method, when refined, can be competitive for larger texts that only need to be decompressed in the real time.