Design And Implementation Of Dissertation Standardization Evaluation System

Dissertation is an important part of graduate education, is a significant symbol of training high quality postgraduate student, and is also the main basis of awarding a degree. As a specific genre of academic paper, a dissertation has a given format specification, normalization is one of the important content of evaluating the quality of a dissertation. At present the normalization inspection of a dissertation relies on manual work and cannot realize automated processing. Processing in this way takes time and energy, and errors are easy to occur. So, in the view of above, we develop an evaluation system of the dissertation normalization.The design scheme of the system is based on the combination of MFC framework and the OLE document processing technology. It achieves that from external interface to read word document, and importing the standard format of the dissertation (the template) and the test dissertation. Then the system will realize automatic detection of the dissertation format and give the result report.First, the system detect the given dissertation format according to the thesis template, including the detection of document format and the detection of chapter sequence and the order detection of graphs tables and formulas labeling. In the detection of reference format module, we propose a method of extracting references name entities based on rule, and the detection algorithm of reference citations order. Then we use the algorithm in the system and realize the normalization detection of references and the order detection of reference citations.Then, the system performs statistical analysis on the basic data of the dissertation, including the recognition of the dissertation structural elements, the number of words and pages in every chapter, the number of words in the dissertation, the number of different words, the number of references, the ratio of Chinese and English references, the time statistics of reference citations etc.Finally, the system evaluates the rationality of the distribution of the parts of the speech. The dissertation uses the natural language statistical processing method to evaluate the distribution rationality; through the dynamic link library, the ratio of nouns, verbs, adjectives, prepositions, adverbs and pronouns appeared in the paper are calculated with the help of the word segmentation system. This is used to assess the preciseness of the organization of the part of the speech; the assessment report is given then.Test results indicate that every function module runs normally and complete the content in the design scheme. Data between each function normally connects and can produce test result report. According to the comparison with the results of manual detection, the system can save detection time and improve efficiency of operation in ensuring the accuracy of detection. Based on the test of format detection in two terms postgraduate dissertations, each function part in the system has been gradually completed. The system can be used in the automatic detection of the dissertation.