A Web-Based Intelligent System for Automated Faculty Document Evaluation

The promotion of faculty in Philippine State Universities and Colleges is guided by National Budget Circular 461 which requires a lot of documentation within certain Key Result Areas. The process of evaluating these documents manually is tedious and error-prone, leading to long administrative delays. To automate this process, this study developed “A Web-Based Intelligent System for Automated Faculty Document Evaluation.†Employing a developmental and experimental design, the researchers trained models on datasets gathered from twenty faculty members. The system employs PyMuPDF and DocTR OCR for text extraction and a hierarchical BERT-based transformer for classification. A hybrid approach is employed for data parsing i.e. Regular Expressions for structured documents and Large Language Models for complex narratives. The results demonstrate BERT as the most suitable model with an accuracy of 98.90% for the Key Result Area and 98.69% for Criterion classification. Granular Sub-Criterion accuracy was achieved at 92% with data augmentation and oversampling. The system is integrated with an interactive dashboard that provides real-time gap analysis and actionable recommendations. This system finally automates the assessment of academic performance while guaranteeing compliance with the institutional guidelines