U.M. Sharif, E.-S. Atlam, W. Hiraishi, K. Morita, M. Fuketa, and J. Aoe (Japan)
Field Association word, Rank, Word attribute
Human can know the subject of document fields by reading only some relevant specific words called Field Association (FA) words. Most of traditional researches focused on how to extract FA words according to scope of fields. All traditional methods extracted FA words depending on frequency without consideration of word attributes. This paper proposes a new technique for determining relevant FA words using the attribute of word. For single FA words, five rank of association are defined. Word attribute describes the characteristics of words such as INTERPRETIVE as well as PERCEPTUAL attributes. Moreover, associated attributes with the word investigate FA words related to relevant fields. From the experimental results, Recall and Precision achieved 83 94% and 72-88% depending on the type of texts. Finally, by comparing the new algorithm to the old one, Recall and Precision are about 15-28% and 11-18% higher than the traditional methods.
Important Links:
Go Back
