Automatic Recognition Of Relation Category Of Non-saturated Compound Sentences With Two Clauses

Chinese Compound Sentences play an important role in the Chinese syntactic,and nearly two-thirds of Chinese sentences are Chinese Compound Sentences.Chinese Compound Sentence is composed of two and more clauses,and clauses are logical semantic relations.Recognition of relation category of compound sentences is the key to the logical semantic relationship,and it is important to the study of Chinese syntactic,semantic analysis and machine translation.However,Automatic recognition of relation category of compound sentences need to take full account of clauses of semantic,context,syntactic characteristics and so on,which are the key and difficult points of logical semantic relations in Chinese Compound Sentences.In the modern Chinese complex sentence,most of Chinese compound sentences have two or three clauses and compound Sentences with three clauses also can be divided into two clauses.Therefore,the study of Chinese compound sentences with multiple clauses is the study of compound sentences with two clauses.Because non-saturated compound sentences does not contain the obvious relation mark,the recognition of the relationship category can not be identified by explicit relation word matching.Non-saturated compound sentences solve mainly the problem,which caused by the implicit relational markings in the logical semantic relations of the preceding and after sentences.The solution of the implied logical semantic relations in Chinese complex sentence not only makes the identification of the relationship category of saturated compound sentences more accurate,but also can provide the theoretical basis for the recognition of the unrelated complex sentence relation category.This article describes a study of non-saturated compound sentences with two clauses,using syntactic theory and the collocation theory of the relation markers of Chinese Compound Sentences.The data source comes from the Corpus of Chinese Compound Sentences and some compound sentences from the search engine.The study finally recognizes relation category of non-saturated compound sentences with two clauses by using a semantic relevancy computation method based on Chinese compound sentence.The research work mainly has the following two aspects:(1)Using the form of saturated and non-saturated compound sentences,relevant theoretical knowledge.This paper constructs the corpus of non-saturated compound sentences with two clauses by using People 's Daily and Changjiang Daily as the data resource,and carries on the statistical analysis to the form of non-saturated compound sentences with two clauses in the corpus.Due to the limited size of the corpus constructed manually,we use search engines to extend corpus resources.Using the existing research results of Chinese complex sentences relation words and the collocation theory of the relation markers,as well as combined with the semantic relevancy computation based on search engine,this article proposes a semantic relevancy computation method based on Chinese compound sentence.The experimental results show that the accuracy of the method is improved greatly compared with the traditional semantic correlation calculation method.At the same time,this method not only takes into account the span,frequency of the core words and the collocation distance of relation words,but also takes into account the semantic information between the clauses,which is helpful for more accurate identification of the relational categories of complex sentences.(2)In the process of recognition of complex sentence relation category,Using hash algorithms and rules to extract relation words.The method 1s more efficient compared with the traditional method.At the same time,we use the recognition rule base of relation words to filter relation word and combine the semantic relevance calculation method based on the Chinese complex sentence to carry on the automatic identification of the relationship category of non-saturated compound sentences with two elauses.The experimental results show that it is more superiority and effectiveness to automatically identify the relationship of non-saturated compound sentences with two clauses by using the semantic relevance calculation method based on the Chinese complex sentence.On the basis of the theory of Chinese complex sentence,this paper uses semantic relevance to identify the relationship of non-saturated compound sentences with two clauses,which not only solve the problem of logical semantics caused by relation words,but also pave the way for the study of compound sentence hierarchy.