Abstract: The selection of indexing words directly affects retrieval results and their precision. As different indexing sources have different text content, the indexing words determined by different sources will be dissimilar. This paper chose respectively different weights of the full text, abstract, starting and ending paragrphs, and key sentences as indexing sources, used automatic indexing algorithm to extract keywords indexing, and then discussed the relation between indexing result and influence factors, such as indexing strategies, weighting of different indexing source, text length and keyword dictionary. The comparative study revealed that re-processing of full-text could improve the indexing quality, while enhanced semantic tool remains another way to improve the result of subject indexing.

Key words: Text indexing, Indexing quality, Indexing strategies, Qualitative comparison