Abstract:Purpose The number of construction dispute cases has maintained a high growth trend in recent years. The effective exploration and management of construction contract risk can directly promote the overall performance of the project life cycle. The miss of clauses may result in a failure to match with standard contracts. If the contract, modified by the owner, omits key clauses, potential disputes may lead to contractors paying substantial compensation. Therefore, the identification of construction project contract missing clauses has heavily relied on the manual review technique, which is inefficient and highly restricted by personnel experience. The existing intelligent means only work for the contract query and storage. It is urgent to raise the level of intelligence for contract clause management. Therefore, this paper aims to propose an intelligent method to detect construction project contract missing clauses based on Natural Language Processing (NLP) and deep learning technology. Design/methodology/approach A complete classification scheme of contract clauses is designed based on NLP. First, construction contract texts are pre-processed and converted from unstructured natural language into structured digital vector form. Following the initial categorization, a multi-label classification of long text construction contract clauses is designed to preliminary identify whether the clause labels are missing. After the multi-label clause missing detection, the authors implement a clause similarity algorithm by creatively integrating the image detection thought, MatchPyramid model, with BERT to identify missing substantial content in the contract clauses. Findings 1,322 construction project contracts were tested. Results showed that the accuracy of multi-label classification could reach 93%, the accuracy of similarity matching can reach 83%, and the recall rate and F1 mean of both can reach more than 0.7. The experimental results verify the feasibility of intelligently detecting contract risk through the NLP-based method to some extent. Originality/value NLP is adept at recognizing textual content and has shown promising results in some contract processing applications. However, the mostly used approaches of its utilization for risk detection in construction contract clauses predominantly are rule-based, which encounter challenges when handling intricate and lengthy engineering contracts. This paper introduces an NLP technique based on deep learning which reduces manual intervention and can autonomously identify and tag types of contractual deficiencies, aligning with the evolving complexities anticipated in future construction contracts. Moreover, this method achieves the recognition of extended contract clause texts. Ultimately, this approach boasts versatility; users simply need to adjust parameters such as segmentation based on language categories to detect omissions in contract clauses of diverse languages.

Annotation and Classification of Relevant Clauses in Terms-and-Conditions Contracts

Unfair clause detection in terms of service across multiple languages

Classification of Contract-Amendment Relationships

AGB-DE: A Corpus for the Automated Legal Assessment of Clauses in German Consumer Contracts

ContractNLI: A Dataset for Document-level Natural Language Inference for Contracts

Detecting Logical Relation In Contract Clauses

LegalPro-BERT: Classification of Legal Provisions by fine-tuning BERT Large Language Model

Identifying Terms and Conditions Important to Consumers using Crowdsourcing

Automated detection of contractual risk clauses from construction specifications using bidirectional encoder representations from transformers (BERT)

Towards Understanding Contracts Grammar: A Large Language Model-Based Extractive Question-Answering Approach

Intelligent detection on construction project contract missing clauses based on deep learning and NLP

A machine learning-based approach to identify unlawful practices in online terms of service: analysis, implementation and evaluation

Towards Mitigating Perceived Unfairness in Contracts from a Non-Legal Stakeholder's Perspective

Unfair TOS: An Automated Approach using Customized BERT

Modeling Legal Reasoning: LM Annotation at the Edge of Human Agreement

Legal information retrieval for understanding statutory terms

Automatic explanation of the classification of Spanish legal judgments in jurisdiction-dependent law categories with tree estimators

Are LLM-based methods good enough for detecting unfair terms of service?

The Law of Large Documents: Understanding the Structure of Legal Contracts Using Visual Cues

Learning from syntax generalizations for automatic semantic annotation

The Right Model for the Job: An Evaluation of Legal Multi-Label Classification Baselines