A Comparative Study of Collocation Extraction Methods from the Perspectives of Vocabulary and Grammar: A Case Study in the Field of Journalism

A Comparative Study of Collocation Extraction Methods from the Perspectives of Vocabulary and Grammar: A Case Study in the Field of Journalism Abstract In all the natural languages that exist in the world, word collocation plays an important role. The methods of collocation extraction are diverse. On this basis, we compare a variety of commonly used collocation methods from the perspective of vocabulary and grammar, so that researchers can choose specific research methods upon specific occasions. This paper selects 16 extraction methods as a comparative analysis strategy. Based on the People's Daily corpus, an in-depth analysis is carried out on the basis of manual labeling accuracy. It is found that in the extraction of Chinese collocations, LMI, PSM, LR, T, F prefer recognition high frequency matching, OR, DICE, J, MS,  2, Z, RR, PMI, P(y|x) and P(x|y) prefer low frequency matching, and the extraction method of high frequency group is better than low frequency group. MI3 can extract both high frequency and low frequency collocations at the same time, and has the characteristics of two groups of methods at the same time, and its accuracy is the highest. Different collocation extraction methods have their own characteristics * Corresponding Author when extracting different syntactic structures. 1 Introduction In all natural languages that exist in the world, word collocations appear frequently and play a pivotal role. In Chinese, because the words lacks and words is not restricted by the language form, so the study of collocation is particularly important． With the rapid development of computer technology, the study of word collocation extraction plays an important role in the field of natural language processing. Collocation extraction refers to the automatic extraction of collocations from a corpus through the computing power of a computer and programming language, Such aspects as machine translation, word sense disambiguation, language generation toand information retrieval are closely related it. Therefore, the research on the automatic extraction method of word collocation is also very important. There are many kinds of collocation extraction methods. Researchers have adopted different collocation extraction Gu Lulu Pan Yue Liu Pengyuan* Department of Information Science, Beijing Language and Culture University Department of Information Science, Beijing Language and Culture University Department of Information Science, Beijing Language and Culture University 15th Xueyuan Road, Haidian District, Beijing 10083, China 15th Xueyuan Road, Haidian District, Beijing 10083, China Center of National Language Recourse and Monitoring Research Print Media Branch 15th Xueyuan Road, Haidian District, Beijing 10083, China gululu_ghrise@126.com viki.pan@qq.com liupengyuan@pku.edu.cn methods in many related studies, and the collocations obtained by different extraction methods are not the same, which will inevitably affect the research conclusions to a certain extent. Some scholars(Xue Jing, Du Youfu, Huang Lan, 2016) conduct comparative studies based on word collocation extraction systems. They mainly compare and analyze commonly used extraction systems from three aspects in order to find a system suitable for specific situations but there is no systematic analysis of the comparison between specific collocation extraction methods. There is also a comparison of extraction methods …

Content truncated to 500 words