Research And Implementation Of Contract Intelligent Review System Based On Natural Language Processing

In commercial transactions,the contract is the legal document that ensures that the rights and interests of the parties are protected.The contract review process is very important to ensuring the legality and validity of the contract.With the rapid development of natural language processing technology,auditing contracts has become one of the important applications of natural language processing technology,bringing new solutions to contract auditing.This paper provides an in-depth analysis and study of contract intelligent auditing in conjunction with natural language processing technology.Firstly,in order to solve the problem of the lack of publicly available Chinese contract text datasets,this paper collected a large number of contract templates on the Internet,and after removing text noise,standardising data formats,correcting content,filling contract templates,developing annotation specifications and annotating the corpus,a high-quality contract annotation corpus was realised,including a corpus of contract clauses and a corpus of named contract entities,which provides support for the next step of the work and also provides the basic conditions for natural language processing work in the contract-oriented domain.Secondly,in view of the long utterances of contract clauses and the influence of keywords on the classification results,this paper designs and implements a multi-model fusion contract clause classification model,combining convolutional neural networks and bidirectional long and short-term memory neural networks to extract key features in the text,and the attention mechanism layer is used to enhance the focus on keywords,which improves the accuracy of the contract clause classification task,and is used in the proposed This paper also presents an experimental validation on the proposed contract clause corpus dataset.Meanwhile,a BERTbased contract named entity recognition model is designed and implemented in this paper to perform named entity recognition on contract text.The model employs a BERT pre-trained model as a word embedding layer,combined with a long and short-term memory neural network and a conditional random field to improve the accuracy of named entity recognition,and is experimentally validated on the contract named entity corpus dataset constructed in this paper.Finally,this paper combines a multi-model fusion contract clause classification model and a BERT-based contract named entity identification model to design and implement a contract intelligent audit system.The main functions of the system include contract upload,contract category classification,contract missing clause detection,contract relative and relationship display.The system has been tested to verify the accuracy of the main functions of the system and to prove that the system can assist in the manual review of contracts and ensure the accuracy of the review.