Abdillah Azmi, Alamsyah Alamsyah

Textual Entailment for Non-Disclosure Agreement Contract Using ALBERT Method

Introduction

Textual entailment for non-disclosure agreement contract using albert method. AI uses ALBERT to analyze Non-Disclosure Agreement (NDA) contracts via textual entailment. Achieves 85% accuracy, outperforming other models. Automate legal text understanding efficiently.

Abstract

Purpose: NDA (Non-Disclosure Agreement) is one type of contract letter. An NDA binds two or more parties who all agree that certain information shared or created by one party is confidential. This type of contract serves to protect sensitive information, maintain patent rights, or control the information shared. Reading and understanding a contract letter is a repetitive, time-consuming, and labor-intensive process. Nevertheless, the activity is still crucial in the business world, as it can bind two or more parties under the law. This problem is perfect for Artificial Intelligence using Deep Learning. Therefore, this research aims to test and develop a pretrained language model that is designed for understanding contract letters through Natural Language Inference task. Method The method used is to train model to perform the language inference task of textual entailment using CNLI (Contract NLI) dataset. ALBERT-base model version that has been tuned to perform textual entailment is used along with LambdaLR for early stopping and AdamW as optimizer. The model is pre-trained with CNLI dataset several times with multiple hyperparameter. Result: As a result, the ALBERT base model that was used showed an accuracy score of 85 and EM score up to 85.04 percent. Although this score is not the State of the Art of the CNLI benchmark, the trained model can outperform other base versions of model that based on BERT and BART, like SpanNLI BERT-base, SCROLLS (BART-base) and Unlimiformer (BART-base). Value: ALBERT is a model that focuses on memory efficiency and small size parameters while maintaining performance. This model is suitable for performing tasks that require long context understanding with minimum hardware requirements. Such a model could be promising for the future of NLP in the legal area.

Review

This paper tackles the significant real-world problem of understanding Non-Disclosure Agreement (NDA) contracts, a task that is vital for businesses but suffers from being repetitive, time-consuming, and labor-intensive. The authors propose an Artificial Intelligence-driven solution, framing contract understanding as a Natural Language Inference (NLI) task using textual entailment. The core aim of this research is to evaluate and develop a pretrained language model capable of efficiently interpreting the complex language found in legal documents, thereby addressing a critical need for automation in the legal domain. The methodology centers on fine-tuning an ALBERT-base model for the textual entailment task using the Contract NLI (CNLI) dataset. The training process incorporated standard optimization techniques such as AdamW and LambdaLR for early stopping, with multiple hyperparameter trials conducted to optimize performance. The results indicate that the ALBERT-base model achieved an accuracy score of 85% and an EM score of 85.04%. While the authors acknowledge that these scores do not represent the current State-of-the-Art on the CNLI benchmark, a notable strength is the model's ability to outperform other base versions of commonly used language models, including SpanNLI BERT-base, SCROLLS (BART-base), and Unlimiformer (BART-base), demonstrating ALBERT's competitive edge within its class. The primary value proposition of this research lies in ALBERT's architectural strengths: its focus on memory efficiency and a smaller parameter count while maintaining robust performance. This makes the model particularly suitable for tasks requiring the analysis of long contexts—common in legal documents—under conditions of minimum hardware requirements. Such characteristics are crucial for practical adoption in the legal industry, where resource constraints can be a significant barrier. This work therefore presents a promising avenue for the future of Natural Language Processing in the legal area, offering an efficient and scalable approach to automating the complex task of contract analysis, potentially paving the way for more accessible AI solutions in legal tech.

Full Text

Comments

You need to be logged in to post a comment.