A Web Search Result Classification System Based on the Degree of Suitability for Specialists.
雅信 鶴田, Yoshiyuki Umemura, Hiroyuki Sakai, Shigeru Masuyama · NTCIR · 2004
We propose a search result classification system based on the degree of suitability for specialists. Our system classifies documents based on whether a document is “for specialists” or “not for specialists”. This classification is done by ranking each document using the density of keywords and by a threshold. For evaluation of our system, we participated in NTCIR4-WEB Topical classification task. Only a part of “evaluation based on distribution of relevant documents” has executed at this time. In this evaluation, our system exhibits good performance for some data set. We also independently conducted experiments for evaluation based on classification errors. Experimental results show that the precision of classification of our system attains value close to that of human evaluators’ average.