カンファレンス (国内) クラウドソーシングを用いたカタカナ文字列の変換について

寺岡 照彦, 坪内 孝太

2016年度人工知能学会全国大会(第30回) (JSAI2023)


This paper proposes a method for converting defective katakana character string to correct Japanese words using a crowdsourcing system. It is difficult to convert a katakana string with missing characters often appeared on credit-card statements due to a character limit. Our methods give hints including Web search resluts of an original katakana string to workers on a crowdsourcing system. Experimental results show our method increases an accuracy of a conversion comparedwith a usual Kana-Kanji conversion.

Paper : クラウドソーシングを用いたカタカナ文字列の変換について新しいタブまたはウィンドウで開く (外部サイト)