06月07日(Fri) 09:00〜11:20 C会場(-国際会議場202号室)
演題番号 | 4C1-IOS-4b-6 |
---|---|
題目 | A machine learning-based approach to missing preposition detection. |
著者 | 大橋 駿介(東京大学大学院情報理工学系研究科コンピュータ科学専攻) 原 忠義(国立情報学研究所) 相澤 彰子(国立情報学研究所 コンテンツ科学研究系,東京大学大学院 情報理工学系研究科 コンピュータ科学専攻) |
時間 | 06月07日(Fri) 10:20〜10:40 |
概要 | Estimated to exceed one billion, the number of those currently studying English as a foreign language is expected to continue growing. Various tools that draw on natural language processing have been developed to help students of English detect and correct writing errors. As one example, tools that correct spelling errors have achieved high accuracy and are now widely used throughout the world. These tools are used not just by those learning English, but in other areas - for natural language processing systems, for instance - to improve output from machine translation systems. Nevertheless, the tools for many other aspects, including grammar checking, remain relatively ineffective. This paper proposes a system for detecting missing prepositions based on syntactic information provided by an English language parser. The information lets us focus on locations that may lack a required preposition. This information can also be used as a machine learning feature to determine whether a location truly requires a preposition. By comparing the detection accuracy achieved against a simple baseline system, our study assessed the effectiveness of our system on the Konan-JIEM Learner Corpus, a typical English learner corpus. We found that our system achieved accuracy superior to the baseline system. |
論文 | PDFファイル |