In spite of the recent advances of information retrieval techniques aiming to serve to search process, the actual reduction of labour of indexing have not been fully accompolished, since traditional automatic derivative indexing techniques, like KWIC/KWOC, which extracts single word as a index term unit from the context, lacks term specificity, and redundancy of index file tends to be large. In this paper we propose a new automatic indexing system, in which index term units are an extracted term string placed between stop-word from the context, named KWEST, an acronym of Key Word Extracted as a String of Terms. KWEST examples of 310 Mathematical Reviews titles show that the reduction of redundancy in comparison with single word KWIC/KWOC system is 47.8%.
View full abstract