Predicting Indian Supreme Court Judgments, Decisions, or Appeals

DOI10.1093/slr/hmac006
Date23 May 2022
Pagesnull-null
Year2022
Published ByOxford University Press
1. INTRODUCTION

Despite several changes coming through Indian jurisprudence such as e-courts,1 the Indian legal system is not expected to be much different arguably from the legacy system and largely to remain inefficient, complex, and quite slow; where in general, the court proceeding time spans expand to several years to decade. The historical evidences suggest, often for several cases, the court processing literally spans over several decades before reaching to the final verdicts.2 The poor percentage of presiding judges versus population size is often being monologued, perhaps since India’s independence, as the primary basis for unbelievably large period of judicial proceedings and final decision making; however, this argument does not better serve the community in any manner, and society at large continues to suffer due to delayed justice system, which then erodes their constitutional rights, which ensure and safeguard fair and faster justice. In addition to several hidden sufferings because of this, the valuable time, energy and resource loss, substantial oneself agony, and critical financial stress, which rapidly inches to indebtedness, are some prominent human states one undergoes through; and consequently, over the period, the yearnings of civil society for swift justice commence to dry and ultimately die.

Therefore, the Indian legal system has enormous space for enhancement and needs immediate focus and collaborative efforts from the interdependent scientific, more specifically computational and legal, communities to find disruptive methods and mechanisms to advance the civil justice system significantly. Lately, computer and legal scientists together have explored numerous possibilities and opportunities of applying Artificial Intelligence (AI), Machine Learning (ML), Deep Learning (DL), and Natural Language Processing (NLP) techniques and technologies into legal system, which has been mostly remained insulated from Information Technology (IT) world.

A few futuristic scientists have taken up the hard task to apply ML and NLP techniques to develop prediction models to accurately predict court decisions with several eventual goals to advance, improve, and elevate the civil justice system; and among others, one prominent goal is to significantly shorten the time period of judicial proceedings and judgment delivery. And in this view, devising and development of efficient and accurate legal predictive models become urgent and of utmost importance to better serve the society. In courts, justices and judges can consider and use prediction models as augmented support to make their decision-making process faster. The lawyers or attorneys can utilize the prediction models to predict case outcome in advance, which will help them to better prepare for defense during court arguments. However, given the complex nature of legal system,3 predicting court judgments and case outcomes with reasonable explanation with greater accuracy is a complex and tedious task and requires intensive general IT, NLP, and ML usage for data—downloading, extracting, cleansing, modeling, and engineering—and model development, training, and testing. Therefore, we also endeavor to contribute to this challenging task and develop an efficient prediction model to closely predict Indian supreme court judgments, decisions, or appeals. And, in this paper, we discuss our newly developed ML-enabled legal prediction model and its operational prototype, called eLegPredict, which successfully predicts the Indian supreme court decisions. The eLegPredict is trained and tested over 3072 supreme court cases and has achieved around 76% accuracy. The eLegPredict is equipped with an interface (a directory) to support and facilitate end users in accepting their new case description as input; the users can drop their case description in pdf format into that designated directory and as soon as the document with new case description is dropped, the system swiftly reads through its content and generates prediction.

The rest of the paper is structured as follows. Section 2 discusses the related work with special focus on court decision predictions. Section 3 is the architecture of eLegPredict. Section 4 discusses the core technical work, the main thrust, of this paper. Section 5 expands on interesting discussion on applied ML and NLP into legal system. And finally, the paper is concluded in Section 6, briefly outlining the future work.

2. RELATED WORK

Katz et al. develop a prediction model to predict the voting pattern for SCOTS.4 Among others, authors also use random forest algorithm with over 60 years SCOTS judgments and claim to achieve 69.7% accuracy on decisions. André Lage-Freitas et al. research and develop a prediction methodology to forecast the judgments from Brazilian court; the author also code and create an operational prototype, which demonstrates around 79% when trained and tested on 4043 Brazilian court cases.5 O’Sullivan and Beel generate ML-enabled models to predict whether ECHR judgments violate any human rights article; to obtain better performance, the authors use word (echr2vec) and paragraph (doc2vec) embeddings and overall 68.83% test accuracy is achieved.6 Pillai and Chandran explore the usage of NLP, with particular emphasis on Bag of Words, and CNN to churn out the important words from Indian court judgments, to classify them—bailable versus non-bailable—and subsequently to predict their classification; their prediction model achieves around 85% accuracy.7 Sert et al. apply AI technique, with NLP’s word embedding, to predict whether the judgments from Turkish constitutional court on individual cases violate any public morality and freedom of expression; the authors claim around 90% accuracy in making correct prediction.8 Sharma et al. apply ML for a very interesting problem, the increasing divorce cases in India; the authors aim to predict the divorce cases, particularly to aid therapists and marriage counselors as the growing divorce number is exerting stress on them; the authors formulate the divorce issue as binary yes/no classification problem and apply their algorithm to predict the divorce possibility with 98.5% accuracy; this algorithm will particularly assist therapists to gauge the gravity of situation between involved parties, and subsequently to strategically plan and guide their counseling efforts in effective direction.9

3. ARCHITECTURE

Figure 1 depicts the complete architecture of eLegPredict. The figure clearly demonstrates the main players, who can take the best advantage of this model—judges, lawyers or attorneys, and the common users with some legal knowledge. The content in existing supreme court judgments consist...

Get this document and AI-powered insights with a free trial of vLex and Vincent AI

Get Started for Free

Start Your Free Trial of vLex and Vincent AI, Your Precision-Engineered Legal Assistant

  • Access comprehensive legal content with no limitations across vLex's unparalleled global legal database

  • Build stronger arguments with verified citations and CERT citator that tracks case history and precedential strength

  • Transform your legal research from hours to minutes with Vincent AI's intelligent search and analysis capabilities

  • Elevate your practice by focusing your expertise where it matters most while Vincent handles the heavy lifting

vLex

Start Your Free Trial of vLex and Vincent AI, Your Precision-Engineered Legal Assistant

  • Access comprehensive legal content with no limitations across vLex's unparalleled global legal database

  • Build stronger arguments with verified citations and CERT citator that tracks case history and precedential strength

  • Transform your legal research from hours to minutes with Vincent AI's intelligent search and analysis capabilities

  • Elevate your practice by focusing your expertise where it matters most while Vincent handles the heavy lifting

vLex

Start Your Free Trial of vLex and Vincent AI, Your Precision-Engineered Legal Assistant

  • Access comprehensive legal content with no limitations across vLex's unparalleled global legal database

  • Build stronger arguments with verified citations and CERT citator that tracks case history and precedential strength

  • Transform your legal research from hours to minutes with Vincent AI's intelligent search and analysis capabilities

  • Elevate your practice by focusing your expertise where it matters most while Vincent handles the heavy lifting

vLex

Start Your Free Trial of vLex and Vincent AI, Your Precision-Engineered Legal Assistant

  • Access comprehensive legal content with no limitations across vLex's unparalleled global legal database

  • Build stronger arguments with verified citations and CERT citator that tracks case history and precedential strength

  • Transform your legal research from hours to minutes with Vincent AI's intelligent search and analysis capabilities

  • Elevate your practice by focusing your expertise where it matters most while Vincent handles the heavy lifting

vLex

Start Your Free Trial of vLex and Vincent AI, Your Precision-Engineered Legal Assistant

  • Access comprehensive legal content with no limitations across vLex's unparalleled global legal database

  • Build stronger arguments with verified citations and CERT citator that tracks case history and precedential strength

  • Transform your legal research from hours to minutes with Vincent AI's intelligent search and analysis capabilities

  • Elevate your practice by focusing your expertise where it matters most while Vincent handles the heavy lifting

vLex

Start Your Free Trial of vLex and Vincent AI, Your Precision-Engineered Legal Assistant

  • Access comprehensive legal content with no limitations across vLex's unparalleled global legal database

  • Build stronger arguments with verified citations and CERT citator that tracks case history and precedential strength

  • Transform your legal research from hours to minutes with Vincent AI's intelligent search and analysis capabilities

  • Elevate your practice by focusing your expertise where it matters most while Vincent handles the heavy lifting

vLex