The Automatic Analysis Method Of Chinese Three-sentence Complex Sentences Based On Deep Learning

The study of Chinese complex sentences belongs to a small category in Chinese information processing.In the field of Chinese information processing,there are many researches on characters and words.However,the study of a complex sentence in Chinese,especially three-sentence complex sentence,is less studied.Compared with simple sentences,complex sentences contain richer semantic information and more diverse forms of expression,so they have higher research value and significance than simple sentences.In the analysis and research tasks of Chinese complex sentences,the analysis and research of two-sentence complex sentence has achieved good results,but the analysis and research of three-sentence complex sentence stands still.One of the important reasons is that there is no corresponding corpus,and the corpus is the foundation of the analysis and research of complex sentences.At present,most of the corpus data used in the study of complex sentences come from the CCCS corpus.Among them,the corpus data is more classic,but is too old.In view of this situation,this article revolves around the establishment of a new corpus-Knowledge Collocation Corpus of relational words in Chinese Complex Sentences.The specific research work includes the following three aspects:First,establishing a corpus"Knowledge Collocation Corpus of relational words in Chinese Complex Sentences" for the study of Chinese complex sentences,marking the three-sentence complex sentences;completing the relational words in these three-sentence complex sentences,and filling them into the full three-sentence complex sentences;simultaneously record the relational words,collocation and combination of the relational words,hierarchical structure and relation categories in the three-sentence complex sentences,to prepare for the subsequent analysis and research tasks;Second,the relationship category in the complex sentence can be determined according to the relationship word,which shows that the relationship word in the complex sentence is important for the study of the complex sentence.In Chinese expressions,the boundaries between words are not very clear,and there are many situations such as polysemy and polysemy,which brings many difficulties to the automatic recognition of relational words.However,with the advancement of technology,the recognition of Chinese complex sentence relational words based on dependency relationship rules and decision tree methods has enabled the recognition rate to reach more than 90%.However,only to identify relation words.In order to facilitate follow-up research,in the task of identifying relational words,in addition to identifying the relational words in the complex sentence,the relation category corresponding to the relational words must also be identified.Therefore,we need to find a new way to better identify the relational words in the complex sentence.In this way,in the research task of this article,we can achieve more results with less effort.In this paper,the method of combining CRF and Bi-LSTM is used to identify the relational words in the complex sentence and the relationship categories corresponding to the relational words,and apply the deep learning method to the relationship word recognition;Third,using the above mentioned method for recognizing relational words in complex sentences,to identify relational words in tripartite complex sentences.According to the relational words,analyze the collocation and combination of relational words to determine whether the three-sentence compound sentence is a full-sentence three-sentence compound sentence;if it is not a full three-sentence compound sentence,then use CNN,Attention and other models to find its default relational words and complement the relation words.Finally,comparative analysis of the experimental results can prove the effectiveness of the research method to a certain extent.