Mathematical Description of Inverted Index File

摘要

Inverted index files are widely used in current information retrieval systems. In inverted index files records of document number an other information are kept one by one. Such storage strategy heavily depends on large RAM and disk I/O, especially the latter, which will easily become the bottleneck. In this paper, a new mathematical description of index files are presented to reduce storage demand and to speedup retrieval procedure. We carried out experiments to test our method. Experiment result shows that the method can reduce storage of frequently appared words and improve retrieve speed at the same time. Stop words are not necessary to be filtered too.