Efficient Clause Identification in Contracts Using NLP and Web-Sourced Data

Kapil Vuthoo, Sonia Khetarpaul, Saurabh Mishra

Abstract: Natural language processing (NLP) is being applied to legal contracts to improve the efficiency and accuracy of the contract review and analysis process. Reviewing contracts is a time-consuming procedure that incurs significant expenses to companies and social inequality to those who cannot afford it. In recent years, the remarkable performance gains enabled by advanced NLP techniques have underscored their value in automating and simplifying work, mainly when applied to legal documents. The primary objective of this paper is to concentrate on clause retrieval within contract documents through the use of general clause data sourced from the Web, as opposed to relying on costly annotated datasets labeled by domain experts. Additionally, the study investigates Question Answering as the baseline for labeling and extends the model through a classification approach. The classifier is trained on clauses extracted directly from diverse publicly accessible clause repositories. Using web-scraped data, this novel technique achieves annotated data generation without requiring manual annotation by legal experts, thereby reducing the overall cost of the dataset creation process.

External IDs:doi:10.1007/978-3-032-02949-2_44