Automatic acquisition of linguistic characteristics and writing support useful for modification of redundant sentences

2014 Fiscal Year Final Research Report

• Project/Area Number: 23500178
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Research Field: Intelligent informatics
• Research Institution: Tottori University
• Principal Investigator: MURATA Masaki 鳥取大学, 工学(系)研究科(研究院), 教授 (50358884)
• Project Period (FY): 2011-04-28 – 2015-03-31
• Keywords: 文生成 / 冗長な表現 / 文の修正 / 機械学習

Outline of Final Research Achievements

In this study, we constructed systems useful for improving redundant sentences. We handled Japanese sentences. Our study contributes to efficiency of communication. We obtained accuracies of about 0.7 to 0.8 in detection of redundant sentences using machine learning. We obtained accuracies of more than 0.6 in correction of redundant sentences including the Japanese words, "kanou" (possible), "toiu" (that is) and "surukoto" (perform), using machine learning. By gathering and examining redundant expressions in long texts including long sentences and long paragraphs, we clarified various kinds of characteristics in redundant expressions of long texts.

Free Research Field

自然言語処理