Thesis Diseño e integración de un motor de reglas flexibles para un algoritmo de planificación: un estudio en la empresa uPlanner
Loading...
Date
2026-02
Authors
Journal Title
Journal ISSN
Volume Title
Program
Ingeniería Civil Informática
Departament
Campus
Campus Casa Central Valparaíso
Abstract
Se diseñó e implementó un prototipo funcional de Motor de Reglas de Negocio (BRE) para el sistema de planificación de exámenes de uPlanner, separando la lógica de restricciones institucionales del algoritmo heurístico mediante un lenguaje específico de dominio (DSL) declarativo basado en JSON. El motor, desarrollado en Python, incluye 18 operadores reutilizables, evaluación individual y grupal, condiciones de aplicabilidad selectiva y un índice de grupos invertido que reduce la complejidad de evaluación en lote de O(n²) a O(n). Validado sobre 4.557 propuestas reales con 24 reglas, alcanzó un 67% de tasa de cumplimiento global, con 10 reglas en 100%, e identificó brechas sistemáticas en distribución de evaluaciones. La solución demuestra la viabilidad de declarar y auditar restricciones académicas sin intervención del equipo de desarrollo.
A functional Business Rules Engine (BRE) prototype was designed and implemented for uPlanner’s exam scheduling system, decoupling institutional constraint logic from the heuristic assignment algorithm through a declarative JSON-based domain-specific language (DSL). The engine, built in Python, features 18 reusable operators, individual and group-scope evaluation, selective applicability conditions, and an inverted group index reducing batch evaluation complexity from O(n²) to O(n). Validated on 4,557 real proposals against 24 rules, it achieved a 67% global compliance rate, with 10 rules reaching 100% compliance, and identified systematic gaps in exam distribution constraints. The solution demonstrates the feasibility of declaratively defining and auditing academic constraints without developer intervention.
A functional Business Rules Engine (BRE) prototype was designed and implemented for uPlanner’s exam scheduling system, decoupling institutional constraint logic from the heuristic assignment algorithm through a declarative JSON-based domain-specific language (DSL). The engine, built in Python, features 18 reusable operators, individual and group-scope evaluation, selective applicability conditions, and an inverted group index reducing batch evaluation complexity from O(n²) to O(n). Validated on 4,557 real proposals against 24 rules, it achieved a 67% global compliance rate, with 10 rules reaching 100% compliance, and identified systematic gaps in exam distribution constraints. The solution demonstrates the feasibility of declaratively defining and auditing academic constraints without developer intervention.
Description
Keywords
Motor de reglas de negocio, Lenguaje específico de dominio, Planificación de exámenes, Restricciones académicas, Arquitectura desacoplada
