@techreport{oai:ipsj.ixsq.nii.ac.jp:00056664,
 author = {勝丸, 真樹 and 駒谷, 和範 and 尾形, 哲也 and 奥乃, 博 and Masaki, Katsumaru and Kazunori, Komatani and Tetsuya, Ogata and Hiroshi, G.Okuno},
 issue = {46(2008-SLP-071)},
 month = {May},
 note = {音声対話システムにおいてユーザはしばしば名称の一部を省略して発話し，音声認識誤りを招く．我々は，「簡略表現」を単語の一部を省略した表現として定義し，簡略表現を音声認識辞書に自動追加する．簡略表現の取得には，日本語では複合語を分割する必要があるが，形態素解析器のみの分割では固有名詞は必ずしも正確に分割できない．さらに，多くの簡略表現を辞書に追加すると，語彙サイズの増加により音声認識精度が劣化する．我々は，前者の問題に対し，形態素解析器に加えて辞書内の文字間の連接確率を用いることで固有名詞の複合語を分割した．後者の問題に対しては，生成した簡略表現の生起確率を，既存辞書の語との音素列の類似度に応じて操作した．本手法で簡略表現を自動追加した結果，既存辞書内の語彙のみを含む発話に対する単語正解精度の劣化を 0.1 ポイントに抑えながら，簡略表現を含む発話の単語正解精度を，既存の辞書による場合と比較して 24.2 ポイント向上させた．, Users often abbreviate long words when using spoken dialogue systems, which results in automatic speech recognition (ASR) errors. We define abbreviated words as sub-words of the original word, and automatically add them into an ASR dictionary. Two issues arise during this vocabulary expansion. The first problem is that proper nouns cannot be correctly segmented by general morphological analyzers, although long and compounded words need to be segmented in agglutinative languages such as Japanese. The second is that, as the vocabulary size increases, adding many abbreviated words degrades the ASR accuracy. We develop two methods, (1) to segment words by using conjunction probabilities between characters, and (2) to manipulate occurrence probabilities of generated abbreviated words on the basis of the phonological similarities between abbreviated and original words. By our method, the ASR accuracy was improved by 24.2 points for utterances containing abbreviated words, with only a 0.1 point degradation of ASR accuracy for those containing words in the original dictionary.},
 title = {音声対話システムにおける簡略表現認識のための誤認識増加を抑制する自動語彙拡張},
 year = {2008}
}