蒙古文真词错误的侦测与纠错建议的生成研究

发布时间：2018-08-12 18:04

【摘要】：本文针对蒙古文文本字词级错误,借鉴其他语言成功的校对系统设计实现了具有真词错误查错与纠错建议生成功能的文本校对系统。在实现该系统的过程中,我们充分利用内蒙古大学蒙古学学院创建的1000万词级"现代蒙古语语料库"。首先,从语料库中提取词语同现频率。其次,在词语同现矩阵的基础上研制了真词错误查错模型。本文主要从以下几个方面进行研究并做了实验。(1)本文实现的蒙古文真词错误校对系统主要有查错和纠错两个模块组成,由于纠错是在查错的基础上进行的下一步操作,因此蒙古文真词错误的查错研究更为重要。首先,我们先利用统计的方法从1000万词级语料库中提取了词语同现频率。然后统计得到的有效词频共1,845,983条。(2)为了对可能出错的词语提供纠错建议,在借鉴人们以往的研究成果的基础上,建立了易混淆词词典,也就是纠错知识库。(3)在词语同现矩阵的基础上,利用词的二元接续关系设计实现了蒙古文真词错误查错算法。并为已查到的真词错误提供纠错建议,在易混淆词词典的基础上设计实现了一种纠错建议生成的算法。(4)在设计实现该系统的基础上,总结出实验全部过程中所遇到的问题及相应的解决方法。再对最后的实验结果进行分析和评价。
[Abstract]:Aiming at word-level errors in Mongolian text, this paper designs and implements a text proofreading system with the functions of error checking and error correction advice generation in other languages. In the process of realizing this system, we make full use of the "Modern Mongolian Corpus" of 10 million words created by the College of Mongolian Studies of Inner Mongolia University. First, the cooccurrence frequency is extracted from the corpus. Secondly, on the basis of word cooccurrence matrix, the model of true word error detection is developed. This paper mainly carries on the research from the following several aspects and has made the experiment. (1) the Mongolian true word error proofreading system realized in this paper mainly has two modules: check the error and correct the error, because the error correction is the next operation on the basis of the error checking. Therefore, it is more important to investigate the errors of Mongolian true words. Firstly, the cooccurrence frequency of words is extracted from 10 million word-level corpus by statistical method. Then, the effective word frequency is 1845983. (2) in order to provide error correction advice for words that may go wrong, a dictionary of confusing words is established on the basis of people's previous research results. That is the error correction knowledge base. (3) on the basis of the cooccurrence matrix of words, the error detection algorithm of Mongolian true words is designed by using the binary continuity of words. An algorithm of error correction is designed and implemented on the basis of confusing word dictionary. (4) based on the design and implementation of the system. The problems encountered in the whole process of experiment and the corresponding solutions are summarized. Then the final experimental results are analyzed and evaluated.
【学位授予单位】：内蒙古大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：H212

【相似文献】