基于深度学习的施工合同纠纷案件法条推荐方法研究

何万磊

摘 要

建筑业作为我国社会经济发展的支柱产业，在高速发展的同时，也催生出了诸如工程质量安全事故、违法发包、工程款拖欠等社会问题，关于施工合同纠纷的案件数量大幅上涨。在进行诉讼前，案件诉讼当事人需要查找案件的适用法条，了解面临的法律问题和法律规定，从而为案件的诉讼提供法律依据。实际操作中，案件诉讼当事人往往选择咨询法律专家（如律师）或选择通过法律搜索系统来查找案件的相关法条。法律咨询十分昂贵，自动法律检索系统是一种更实惠的法条查找方式。当前主流的法律检索系统虽然提供了法条的查询方法，但用户无法通过简单的案件陈述获得相关的法条，需要专业的法律知识。因此，需要一个专业的施工合同纠纷案件法律查询系统，允许诉讼当事人通过输入自然语言描述案件事实快速查找案件的适用法条。

本文提出了一种基于深度学习的施工合同纠纷案件法条推荐方法。首先通过对施工合同纠纷判决书的内容进行分析，利用文本挖掘技术，构建了施工合同纠纷案件法条推荐数据集，共包含11481条有效数据。其次，通过对施工合同纠纷案件法条推荐任务进行分析，将法条推荐任务转换为法条多标签分类问题，引入sigmoid交叉熵损失，构建TextCNN多标签分类模型，并在施工合同纠纷案件法条推荐测试集上对比了多种常用多标签分类方法，结果表明TextCNN多标签分类模型在测试集上的表现最好。最后，设计了施工合同纠纷案件法条推荐原型系统，通过结合TextCNN多标签分类模型和挖掘的法条关联规则，对查询的案件适用法条进行共现和关联扩展，帮助诉讼当事人查找其涉及的施工纠纷合同案件的相关法条。

本文提出的案件适用法条推荐研究，有助于诉讼当事人通过自然语言描述案件事实快速查找案件的适用法律。本文研究成果对使用大数据和人工智能技术解决施工合同纠纷有借鉴意义。

关键词：施工合同纠纷案件；法条推荐；深度学习；TextCNN；关联规则

Abstract

As one of the pillars industry of China's social and economic development, the construction industry has spawned social problems such as engineering quality and safety accidents, illegal contracting, and arrears of engineering payment while developing at a high speed, resulting in a sharp increase in the number of cases involving construction contract disputes. Before starting litigation, the litigant needs to search the applicable statutes of the litigation cases so they can understand the legal issues and legal regulations to provide a legal basis for the litigation case. When people have legal issues, it is critical to know which statutes are involved. Usually, because of a lack of legal knowledge, they may seek help from legal experts, such as attorneys, or from automatic systems. Since legal consultation is very costly, automatic systems are a much more affordable form of legal support. Although the current mainstream automatic systems provide query methods, users cannot obtain pertinent statutes through simple case statements, requiring professional legal knowledge. Therefore, This motivated us to propose a professional statutes retrieval system that allows litigants to quickly find the applicable statute of the case by entering the natural language to describe the facts of the case.

In view of the above problems in the litigation process of construction contract disputes, this paper proposes a statutes recommendation method for construction contract disputes case based on deep learning. Firstly, based on the analysis of the content of the judgment of construction contract dispute, the dataset applied to statutes recommendation task of construction contract dispute case is constructed by using text mining technology, which contains a total of 11481 valid data. Secondly, by analyzing the statute recommendation task in construction contract dispute cases, the task is converted into a multi-label classification problem of legal articles. By adding Sigmoid cross-entropy losses to the traditional TextCNN multi-classification model, the TextCNN multi-label classifier is constructed as a statutes recommendation method. Finally, a prototype system for recommending statutes in construction contract dispute cases is designed, which combines the TextCNN multi-label classification model and mining statute association rules that are used to extend the applicable statute of the queried cases. The system is convenient for litigants to find relevant statutes for their contract cases involving disputes.

In this paper, research on the statute recommendation is helpful for litigants to find applicable statutes quickly by entering natural language to describe the facts of the case. The research result can be used for reference to solve construction contract disputes by using big data and artificial intelligence technology.

Keywords: Construction Contract Disputes Case, Statute Recommendation, Deep Learning, TextCNN, Association Rules Mining