CapRe: An AI-Driven Document Processing and Recommendation System Using Semantic Embeddings for Automated Research Management
Year: 2026
Type: conference-paper
Abstract: The rapid growth of school research collections creates a need for smarter systems to search, check, and manage academic work. Many current school repositories use simple keyword searches. This often leads to weak results, repeated topics, and long manual checking. To help solve this, the present paper introduces CapRe, a multi-level AI system designed to make finding, reviewing, and managing capstone projects easier. CapRe uses an approach aligned with modern information-retrieval and semantic search technology. This study employed a mixed-method approach, evaluating the system by collecting and analyzing quantitative and qualitative feedback from 182 students and faculty users. The evaluation showed high user satisfaction, with ease-of-use ratings averaging 3.98/5.00 and a strong willingness to recommend the system (91.1%). A key finding is that 91.1% of users indicated they would recommend CapRe to others, showing strong acceptance. The system uses the all-mpnet-base-v2 model to understand the meaning of long research papers, turning documents into numerical vectors for storage and search. When a user uploads a new file, CapRe creates an updated record of its meaning, making future searches more accurate. Results from the users showed that the most helpful features were the file tracking system, the search and filter tools, and the AI topic suggestions. Users reported that CapRe helped them see similar studies, avoid repeated topics, and save time. While some users noted slow loading on mobile, the overall response was very positive. In conclusion, CapRe demonstrates that AI can support better research decisions, clearer topic refinement, and more organized research management in schools.
Institution: University of Cebu-Banilad
Language: English
Subfield: Information Systems
Field: Computer Science
Domain: Physical Sciences
SDG: Quality education
Open Access status: hybrid
OpenAlex ID: W7204532763