131 巻 (2011) 10 号 p. 1819-1827
Recently, digital documents in companies increase rapidly and are required to be ensured. Pop-up of business information according to PC usage situation can achieve this goal. Two functions are required to it: One is to correctly discriminate pop-up conditions from contents of business-use application, and the other is to select suppression keywords from incorrect pop-up documents. The suppression keywords should improve true negative rate without drop of recall. Based on observation of typical phrases and increase of frequencies for characteristic phrases in contents of business-use applications, we select suppression keywords according to their frequency in positive and negative cases and their occurrence in feedback document. Our evaluation using three kinds of business information shows that the proposed method can achieve both high true negative rate and high recall, and to correctly discriminate pop-up conditions.
J-STAGEがリニューアルされました! https://www.jstage.jst.go.jp/browse/-char/ja/