• Ajita Mishra
• Lakshmi Dubey
• Pravishi Gupta
• Akanksha Verma
• Yusuf Perwej

District and taluka courts in India are struggling with a persistent backlog of cases, and they do not have a technique that is data-driven and methodical for determining the priority of cases or grouping things that are related. The purpose of this paper is to present an artificial intelligence-based judicial workload optimization system, which is internally referred to as Nyaya Setu. This full-stack application addresses three coupled sub-problems within the realm of judicial triage. These problems include automatically grouping cases that are similar, generating a transparent priority-ranked cause list, and load-balancing case assignment across three judges. After iterative feature engineering and log-transformation of the skewed target, the system achieves a mean absolute error of 668.5 days by combining a pretrained transformer (BART) for summarization, TF-IDF/K-Means clustering for grouping similar cases, and a LightGBM regression model trained on 3,179 real disposal-duration records to estimate case urgency. This results in the system achieving a mean absolute error of 668.5 days. The system was validated end-to-end on 854 real Supreme Court judgments (2023) with full text: clustering produced five legally interpretable groups (e.g., bail/criminal matters, land-acquisition disputes), and feature-importance analysis of the trained priority model confirmed that case type and state jointly dominate predicted disposal time, consistent with independently published regional disposal-time disparities. We give these findings, which include the significant absolute inaccuracy of the model, with explicit critical analysis rather than as a final solution. Additionally, we suggest clear future approaches, which include deeper per-case characteristics and constraint-based judge assignment.

Judicial Case Management, Natural Language Processing, Text Summarization, Unsupervised Clustering, TF-IDF, Gradient Boosting, LightGBM, Workload Optimization

"AI-Driven Predictive Judicial Workload Management System for Intelligent Case Scheduling, Resource Allocation, and Court Performance Optimization", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.13, Issue 10, page no.a257-a265, October-2026, Available :http://www.jetir.org/papers/JETIR2610027.pdf

"AI-Driven Predictive Judicial Workload Management System for Intelligent Case Scheduling, Resource Allocation, and Court Performance Optimization", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.13, Issue 10, page no. ppa257-a265, October-2026, Available at : http://www.jetir.org/papers/JETIR2610027.pdf