Font Size: a A A

Studies Of Error Detection And Recognization Based On The Usage Of Function Words

Posted on:2014-01-12Degree:MasterType:Thesis
Country:ChinaCandidate:M J LiangFull Text:PDF
GTID:2248330398978331Subject:Computer software and theory
Abstract/Summary:PDF Full Text Request
Chinese function words, which are rich and varied, do not have morphological markers and inflections,however,they play significant roles in syntax and semantics, therefore, they are more flexible and difficult to grasp, hence it is of great importance to study Chinese function words. Automatic identification of modern Chinese function words is based on knowledge base of the usage of function words. The bigger and ampler knowledge base of the usage of function words is, the better the automatic identification of the usage of Modern Chinese function words benefits. The paper briefly introduces the concepts of knowledge base of the usage of function words and the frame of "trinity",and introduces the usage of the modern Chinese function words knowledge base in detail. It includes three parts:modern Chinese function words usage dictionary, modern Chinese function words usage rule base and modern Chinese usage corpus.There are two basic ways for the automatic identification of function words: rule-based method and statistic-based method, as well as the method that combining them. The paper uses rule-based approach to describe the automatic identification of function words usage. Since a function word has different meanings in a variety of ways, the identification rules are various. Using different rule orders in automatic identification process of function words usage can result in different usage recognition accuracy. Total ordering of the rules and making labels on them can be the best way to order. However, the time complexity is high. Screening the result of total ordering first and making labels afterwards will greatly reduce the time complexity.The word "error" is mainly used for inter-language errors in the field of Second Language Acquisition, wrong sentences of middle school students in exams has different sentence patterns and characteristics of the errors from "error",but they could both be divided into four kinds:wrong sequence, improper addition, overrepresentation and improper omission,and the misuse of function words takes a large proportion in each one. Hence it is feasible to use the same method which is based on the usage of function words to make automatic identification study on it. In general, the errors can be divided into four kinds.However,it is complicate to study various usages of specific function word. From a respective of the usage of function words,this paper uses rule-based method and focuses on parts of the four kinds of errors, and accuracies are83.67%、91.56%、87.75%、93.74%. The experimental result shows that, this method can effectively identify the usage error of function words.
Keywords/Search Tags:The usage of function words, Rules, Sorting, Error, Automatic annotation
PDF Full Text Request
Related items