Part-of-Speech Filtering Method in Information Retrieval

• Main Authors: Kai-Wen Yang, 楊凱雯
• Other Authors: Guo-Wei Bian
• Format: Others
• Language: zh-TW
• Published: 2009
• Online Access: http://ndltd.ncl.edu.tw/handle/79541714918077090492

碩士 === 華梵大學 === 資訊管理學系碩士班 === 97 === In this paper, a word-based indexing method with part-of-speech filtering is proposed for information retrieval. Three different filtering methods: noun, noun-verb and noun-verb-adjective-adverb are used to compare with the traditional word-based method and the bi-gram method. The experimental results show that the gap of the performance between the bigram method and the word-based method is small. In most cases, the performances of the word-based method with part-of-speech filtering are better than the bigram method. Compare to word-based method, the word-based method with part-of-speech filtering raises obviously the retrieval performances for both of the Headline field and all fields. But the word-based method with part-of-speech filtering shows 5% improvement in average for the retrieval of TEXT field. The word-based method and the word-based method with part-of-speech filtering have smaller indices and less retrieval time than the bigram method. Both methods are more suitable for the very-large scale information retrieval.