Research On Intelligent Detection The Risk Of Construction Project Contract Missing Clauses Based On NLP And Deep Learning

At present,the new contracting model and consulting model represented by EPC,PPP general contract and whole process consulting contract have achieved a wider range of applications.Generally project contractors and consulting units undertake many engineering projects and sign different types of contracts with multiple contractors at the same time.This situation brings a lot of contract risk management work.However,currently construction project contract risk identification relies on manual review for a long time,which is inefficient and easily restricted by personnel experience.The level of informatization of contract risk management needs to be improved.Therefore,this study used Natural Language Processing(NLP)and Deep Learning(DL)technology to design an algorithm that uses intelligent methods to detect construction project contracts potential missing clauses risks.Futher more,the study provided a feasible technology for improving the efficiency of contract risk management and getting rid of the inherent mode of contract management.Firstly,this study designed a complete scheme of detecting contract missing clauses risk,and introduced the linguistic theory in NLP into the research of construction project contract risk management.The study used Web Crawlers and Easy Data Augmentation technology to obtain a large amount of public contract corpus,and then used Chinese Word Segmentation,Stop Word Removal,Text Representation and other NLP technologies to transform text corpus into a language form that could be understood by computer.Moreover,the study used Convolutional Neural Network and Long-short-term Memory Neural Network to learn feature mapping of contract text to automatically extract contract text features,and fused two models to improve the performance of classification model.Finally,the study researched the similarity algorithm of construction project contract clauses and designed a multi-dimensional labeling system of contract clauses by taxonomy.Based on the BERT model,the study designed a multi-label classification algorithm of contract clauses to reduce the amount of semantic similarity data calculation,and then introduced MatchPyramid structure into BERT to design the contract clauses semantic similarity algorithm.This algorithm could obtain the word feature matching matrix and perform convolution operation,calculating the clauses semantic similarity according to the text interaction information.On the basis of above research,the study designed a system of construction project contract missing clauses detection,and verified algorithm and analyzed effectiveness on the system.The results showed that the NLP and Deep Learning-based intelligent detection method of construction project contracts potential missing clauses risks in this paper could not only identify different contract types of construction such as construction,prospecting,supervision,etc,but also realize automatic label clauses and match text.This method provides intelligent decision-making assistance for the traditional contract risk management relying on manual review and provides a feasible ideas and approaches for the long-standing contract risk management problems such as inconsistent understanding of clauses,inefficient review and incomplete review.