Sample of Legal Provision Retrieval Using Topic Clustering to Support Utility Relocation Agreement Drafting in Highway Projects

Abstract

Utility agreements are legal documents containing specialized clauses that legally define stakeholders’ responsibilities for eliminating utility relocation delays in the right of way of highway projects. However, retrieving sample language for developing such an agreement remains a manual, time-consuming, and labor-intensive task. No previous study has developed a database for quick retrieval of sample clauses retrieval. This paper presents the first effort to develop a well-structured database of utility-related clauses in highway construction projects. This is achieved through an unsupervised topic modeling that utilizes latent Dirichlet allocation (LDA) and natural language processing (NLP) techniques. Through this model, utility-related legal clauses were categorized into six prominent topics. Topics such as dispute resolution and communication emerge as primary areas, encompassing a substantial number of clauses. By harnessing the LDA-based topic classification model, transportation departments can effectively identify prevailing topics and patterns in legal language used by other agencies. This advancement streamlines the process of formulating provisions concerning utility relocation, thereby enhancing the overall efficiency of highway projects.