<![CDATA[Problems in lost and found management within campus environments, which predominantly rely on manual processes or passive digital portals, often result in low recovery rates. Conventional systems fail due to exact keyword-matching limitations, especially when terminology differences exist between reporters and security officers. This research proposes an AI-integrated web solution to overcome these inefficiencies. Utilizing the Gemini Vision API, the system autonomously extracts visual attributes from photos of found items into text descriptions. Furthermore, a Hybrid Matching algorithm combining Jaccard Index and Cosine Similarity computations via a Natural Language Processing (NLP) model evaluates semantic closeness for intelligent match detection. Developed using a qualitative approach involving direct observation and literature review, the software applies the iterative Agile Personal Extreme Programming (PXP) method. This allows gradual calibration of AI probability scores across planning, design, implementation, and testing stages. The resulting secure application dynamically integrates inventory management, presenting proactive match recommendations, cross-table status updates, and automated email notifications. In conclusion, this system acts as an analytical agent, assisting security officers in resolving cases proactively, structurally, and efficiently.]]>
Copyrights © 2026
