本文關(guān)鍵詞：基于循環(huán)神經(jīng)網(wǎng)絡(luò)的蒙古文語言模型研究　出處：《內(nèi)蒙古大學(xué)》2017年碩士論文　論文類型：學(xué)位論文

  更多相關(guān)文章： 語言模型 中間字符 語音識(shí)別 N-gram FRNNLM

【摘要】：語言模型是自然語言處理任務(wù)中的重要組成部分。其中,N-gram語言模型是目前應(yīng)用最為廣泛的統(tǒng)計(jì)語言模型。近年來,隨著深度學(xué)習(xí)技術(shù)的不斷發(fā)展,深度神經(jīng)網(wǎng)絡(luò)模型逐漸被應(yīng)用于語音識(shí)別中,它為研究者帶來新一輪的研究熱潮。神經(jīng)網(wǎng)絡(luò)語言模型是其中比較重要的研究方向之一。蒙古文語言模型對(duì)于蒙古語語音識(shí)別、蒙古文信息檢索和蒙古文機(jī)器翻譯等蒙古文信息處理技術(shù)的研究起著至關(guān)重要的作用�，F(xiàn)階段,神經(jīng)網(wǎng)絡(luò)語言模型已被廣泛應(yīng)用于英文和漢文中,但是神經(jīng)網(wǎng)絡(luò)語言模型在蒙古文中的使用還比較少。本文主要針對(duì)蒙古文神經(jīng)網(wǎng)絡(luò)語言模型進(jìn)行研究。蒙古文是一種在國(guó)際上有廣泛影響的語言文字。然而,在蒙古文文本語料中,存在大量的顯現(xiàn)形式相同但編碼不同的單詞,這給蒙古文單詞的統(tǒng)計(jì)和檢索等帶來了很大困難。本文著重解決顯現(xiàn)形式相同但編碼不同蒙古文單詞的統(tǒng)計(jì)和檢索問題,從而提高蒙古文語言模型的性能。首先,提出了采用中間字符對(duì)蒙古文顯現(xiàn)形式相同但編碼不同的字母進(jìn)行合并表示的方法;接著,分別建立了基于拉丁字符的N-gram語言模型與基于中間字符的N-gram語言模型,以及基于拉丁字符的快速循環(huán)神經(jīng)網(wǎng)絡(luò)語言模型(Faster Recurrent Neural Network Language Model,FRNNLM)與基于中間字符的FRNNLM;然后,實(shí)現(xiàn)了 N-gram語言模型和FRNNLM融合的方法,得到了性能更好的語言模型;最后,用困惑度評(píng)價(jià)了蒙古文語言模型的性能,并將其應(yīng)用到蒙古語語音識(shí)別中進(jìn)行詞錯(cuò)誤率(Word ErrorRate,WER)的比較。實(shí)驗(yàn)結(jié)果表明,基于中間字符的蒙古文文本語料的詞匯量比基于拉丁字符的語料平均減少了 41%;基于中間字符的語言模型(3-gram、FRNNLM)比相應(yīng)基于拉丁字符的語言模型在困惑度方面下降了近40%,提高了蒙古文語言模型的性能。并且在蒙古語語音識(shí)別中,基于中間字符的語言模型(3-gram、FRNNLM、3-gram+FRNNLM)比相應(yīng)基于拉丁字符的語言模型在WER方面下降了近20%;3-gram+FRNNLM(基于拉丁字符、基于中間字符)比3-gram、FRNNLM在WER方面下降得更加明顯,有效提升了蒙古語語音識(shí)別的準(zhǔn)確率。
[Abstract]:Language model is an important part of natural language processing. N-gram language model is the most widely used statistical language model. In recent years, with the development of in-depth learning technology. Depth neural network model is gradually used in speech recognition. It brings a new wave of research for researchers. Neural network language model is one of the more important research directions. Mongolian language model for Mongolian speech recognition. The research of Mongolian information processing technology such as Mongolian information retrieval and Mongolian machine translation plays an important role. At present, neural network language model has been widely used in English and Chinese. However, the use of neural network language model in Mongolian language is relatively small. This paper mainly focuses on the Mongolian neural network language model. Mongolian language is a kind of language which has a wide influence in the world. However. In the Mongolian text corpus, there are a large number of words with the same manifestation but different codes. This brings great difficulties to the statistics and retrieval of Mongolian words. This paper focuses on solving the problem of statistics and retrieval of Mongolian words with the same manifestation but different codes. In order to improve the performance of Mongolian language model. Firstly, the method of combining middle characters to represent Mongolian characters with the same form but different encoding is put forward. Then, the N-gram language model based on Latin character and N-gram language model based on intermediate character are established respectively. And the fast loop neural network language model based on Latin characters, Recurrent Neural Network Language Model. FRNNLM) and FRNNLM based on intermediate characters; Then, the N-gram language model and the FRNNLM fusion method are implemented, and a better performance language model is obtained. Finally, the performance of the Mongolian language model is evaluated with the degree of confusion, and the word ErrorRate is applied to the Mongolian speech recognition. The experimental results show that the vocabulary of Mongolian text corpus based on intermediate characters is 41% less than that based on Latin characters; The language model based on intermediate characters / FRNNLM) is nearly 40% less confusing than the corresponding language model based on Latin characters. The performance of Mongolian language model is improved, and in Mongolian speech recognition, the language model based on intermediate character is 3-gram/ FRNLM. 3-gram FRNNLM) is nearly 20% lower in WER than the corresponding Latin character-based language model; 3-gram FRNNLM (based on Latin characters, based on intermediate characters) is significantly lower in WER than 3-gram FRNNLM. It effectively improves the accuracy of Mongolian speech recognition.
【學(xué)位授予單位】：內(nèi)蒙古大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2017
【分類號(hào)】：TP391.1;TP183

【參考文獻(xiàn)】

相關(guān)期刊論文 前6條

1 張劍;屈丹;李真;;基于循環(huán)神經(jīng)網(wǎng)絡(luò)語言模型的N-best重打分算法[J];數(shù)據(jù)采集與處理;2016年02期

2 王龍;楊俊安;陳雷;林偉;;基于循環(huán)神經(jīng)網(wǎng)絡(luò)的漢語語言模型建模方法[J];聲學(xué)技術(shù);2015年05期

3 王龍;楊俊安;陳雷;林偉;劉輝;;基于循環(huán)神經(jīng)網(wǎng)絡(luò)的漢語語言模型并行優(yōu)化算法[J];應(yīng)用科學(xué)學(xué)報(bào);2015年03期

4 蘇傳捷;侯宏旭;楊萍;員華瑞;;基于統(tǒng)計(jì)翻譯框架的蒙古文自動(dòng)拼寫校對(duì)方法[J];中文信息學(xué)報(bào);2013年06期

5 斯·勞格勞;;基于不確定有限自動(dòng)機(jī)的蒙古文校對(duì)算法[J];中文信息學(xué)報(bào);2009年06期

6 徐望,王炳錫;N-gram語言模型中的插值平滑技術(shù)研究[J];信息工程大學(xué)學(xué)報(bào);2002年04期

相關(guān)碩士學(xué)位論文 前4條

1 江布勒;基于規(guī)則的蒙古文自動(dòng)校對(duì)方法研究[D];內(nèi)蒙古大學(xué);2014年

2 張劍;連續(xù)語音識(shí)別中的循環(huán)神經(jīng)網(wǎng)絡(luò)語言模型技術(shù)研究[D];解放軍信息工程大學(xué);2014年

3 飛龍;蒙古語語音識(shí)別系統(tǒng)的研究與優(yōu)化[D];內(nèi)蒙古大學(xué);2009年

4 趙軍;基于音節(jié)統(tǒng)計(jì)語言模型蒙古文詞匯分析校正器的設(shè)計(jì)與實(shí)現(xiàn)[D];內(nèi)蒙古大學(xué);2007年

，

本文編號(hào)：1441211