Banca de QUALIFICAÇÃO: DENIS RENATO RAMALHO OROZCO

Uma banca de QUALIFICAÇÃO de MESTRADO foi cadastrada pelo programa.
STUDENT : DENIS RENATO RAMALHO OROZCO
DATE: 31/03/2026
TIME: 16:00
LOCAL: Google Meet
TITLE:

Explainable Analysis of Complexity and Duration in Labor Lawsuits: A Data-Driven Approach

KEY WORDS:

Labor Justice; machine learning; explainable artificial intelligence; procedural complexity; judicial efficiency; jurimetrics

PAGES: 107
BIG AREA: Ciências Exatas e da Terra
AREA: Ciência da Computação
SUMMARY:

Judicial delays are a recurring problem in the Brazilian judicial system and directly affect institutional efficiency, access to justice, and the parties’ right to a decision within a reasonable time. In the Labor Court system, this challenge is intensified by the high volume of cases and by case allocation practices that do not take into account relevant differences in case complexity. In this context, the objective of this dissertation is to identify and analyze which characteristics of labor cases in the state of Paraíba are most strongly associated with case processing time, seeking to understand recurring patterns that may support judicial management and contribute to a more balanced allocation of institutional resources. Regarding methodology, this study analyzed 155,312 first-instance labor cases closed between 2013 and 2024, based on a broad set of procedural, factual, and institutional variables. Following an extensive literature review and exploratory data analysis, supervised machine learning techniques were applied—particularly regression models—not with the purpose of predicting the duration of new cases, but as an analytical tool to model complex associative relationships present in historical data. To interpret model behavior, Explainable Artificial Intelligence (XAI) techniques were employed, with emphasis on the SHAP method, used to estimate the marginal contribution of variables to the model outputs. The results indicate that the duration of labor cases is associated with a set of characteristics frequently observed in the analyzed data. It was found that more complex cases, those involving a greater number of procedural stages, cases handled by judicial units with higher workloads, or cases not fully processed in digital format tend to be associated with longer processing times. These findings reveal consistent statistical patterns in the analyzed dataset. It is emphasized that such associations describe recurring behaviors observed in historical data and do not allow for the inference of cause-and-effect relationships. By prioritizing model explainability and the understanding of patterns associated with case duration, this study contributes to the development of more informed indicators of procedural complexity, provides empirical support for judicial management, and reinforces the use of data-driven approaches to improve the Labor Court system, in line with the United Nations Sustainable Development Goals.

COMMITTEE MEMBERS:
Presidente - 2353000 - ELIAS JACOB DE MENEZES NETO
Interno - 2668551 - ANDRE MORAIS GURGEL
Interno - 1669545 - DANIEL SABINO AMORIM DE ARAUJO