Construction of Semantic Collocation Bank Based on Semantic Dependency Parsing

Collocation has always been an important issue in language research, especially in Chinese language researches. Chinese is an isolated language, which lacks morphological changes.Establishing a relatively complete dictionary of Chinese collocation will be a great contribution to Chinese study and research. Collocation plays a significant supporting role in many fields of NLP, such as information retrieval, machine translation, information extraction, and so on. Ding and Bai proposed a method of query expansion based on local co-occurrence [1] ; Lin put relationship ofcollocation into language model for query expansion, which got over the deficiency of insufficient relationShips caused by lacking context in tradition query [2] . In the basic research field of NLP, such as syntax, semantics, etc., collocation also plays an important role.Based on the comparison of different patterns in adjective collocation between the Chinese English learners and native speakers, Zhang analyzed the typical characteristics of different learners when using adjective collocations [3] ; Xingemphasized on the importance of collocation in the second language learning [4] . The early research of automatic collocation extraction was made by Choueka, Klein and Neuwtiz,they defined collocation as adjacent words, and used co-occurrence frequency to extract collocation [6] ;Church and Hanks improved the automatic extraction technology and put forward mutual information as the index ofcollocationevaluating [7] .By proposing a formula for calculating strengthbetweencollocation,introducing dispersion formula,as well as integrating with the automatic speech tagging technology, the Xtract system of Smadja improved the extraction accuracy rate of collocation extraction up to 80% [8] ; Lin extracted collocation based on shallow syntactic parsing [9] ;Shouxun YANG applied the method of decision tree to extract collocation by integrating frequency, likelihood ratio, point mutual information, variance and other statistical indicators [10] . In China, there werea number of outstanding dictionaries had been published, PACLIC 29