CTU Daanbantayan Campus Library · Capstone Repository

About RE-CAPS

A next-generation digital undergraduate theses repository designed for Cebu Technological University — Daanbantayan Campus. Powered by Pinecone vector semantic search, a 4-tier cascading AI research engine, and instant multi-format academic citation generation.

0%
Digital Indexing
Past physical binders digitized and cataloged
0-D
Dense Vectors
Multilingual E5 embedding space via Pinecone
0
Cascading AI Engine
Mistral → Groq → Gemini → OpenRouter
0 Styles
Citation Generator
APA 7th, IEEE, MLA, Chicago, Harvard, BibTeX
The Academic Challenge
Why Traditional Research Archives Fall Short
Before RE-CAPS, undergraduate researchers and faculty advisers at CTU Daanbantayan faced systemic hurdles when trying to discover, evaluate, and cite past capstone manuscripts.
Physical Binder Silos
Decades of capstones locked in shelf binders with no digital index, accessible only during library hours.
Keyword Search Blindness
Traditional exact keyword searches miss relevant studies that use conceptual synonyms, acronyms, or differing academic phrasing.
Unintentional Duplication
Students frequently proposed capstone ideas that had already been defended in prior semesters due to lack of cross-year visibility.
Citation Overhead
Manual bibliography formatting in APA, IEEE, or MLA caused citation inaccuracies and wasted critical writing time.
Architecture Breakdown
How the RE-CAPS RAG Pipeline Works
Retrieval-Augmented Generation (RAG) grounds language models in verified university thesis data, eliminating AI hallucinations and providing source citations.
Stage 01
Vector Ingestion
Each thesis title, abstract, methodology, and key findings are transformed into 1024-dimensional dense vectors using the multilingual-e5-large model.
multilingual-e5-large
Stage 02
Pinecone Similarity
When a student searches, their natural language query is vectorized in real time and matched against the Pinecone index using cosine similarity (80%+ threshold).
Pinecone Vector DB
Stage 03
Context Grounding
The top relevant manuscripts, author credits, advisers, and key discoveries are injected into the AI context window as authoritative ground truth.
System Prompt Grounding
Stage 04
Cascading Synthesis
A 4-tier cascading engine (Mistral → Groq → Gemini → OpenRouter) streams word-by-word academic responses with real citations via Server-Sent Events.
SSE Streaming API
Interactive RAG Query Simulator
Click a sample research query below to preview simulated vector matching and RAG synthesis:
Simulated Query:
"IoT-based Smart Aquaculture & Water Quality Monitoring in Northern Cebu"
Pinecone Cosine Match:
97.8%
IoT-Driven Water Quality Monitoring and Automated Aeration for Coastal Aquaculture in Northern Cebu
BSIT A.Y. 2025 Adviser: Prof. M. Arriesgado

Utilizes an ESP32 sensor cluster measuring dissolved oxygen (DO), pH, and salinity with MQTT telemetry and cloud alarms. Evaluated at Daanbantayan coastal fish ponds, the system demonstrated a 34% reduction in unexpected aquatic mortality and enabled real-time mobile notifications for farm operators.

RAG AI Synthesis & Academic Attribution:
DO levels remained consistently above 5.0 mg/L via automated relay activation; average notification latency was 1.4 seconds.
System Capabilities
Built for Real Academic Discovery
RE-CAPS is not just a digital catalog — it is a complete, production-grade research platform built with modern cloud and AI infrastructure.
Dual Search Engine
Instant Boolean filtering by year, program, and adviser combined with Pinecone vector semantic search for contextual queries.
4-Tier Cascading AI
Resilient multi-model fallback across Mistral, Groq, Google Gemini, and OpenRouter, streaming answers with real-time SSE.
Cloudinary Visual Gallery
High-resolution figure showcases, architecture diagrams, and hardware photos hosted with dynamic Cloudinary transformations.
Cloud & Guest Bookmarks
Save theses locally as a guest, then automatically merge your reading collection into Firestore when creating an account.
Role-Based Access Control
Tailored for Every Campus Stakeholder
RE-CAPS serves four distinct user groups with dedicated workflows, permissions, and specialized dashboards.
Undergraduate Researchers
Student Researcher
Students can search the repository using traditional filters or the AI-powered Pinecone vector search, read abstracts and key findings, save bookmarks, and export formatted citations.
  • Natural-language semantic search via Pinecone Vector Database
  • Conversational Q&A with the RAG Assistant grounded in CTU theses
  • Instant multi-format citation export (APA 7th, IEEE, MLA, Chicago, BibTeX)
  • Personal Cloud Bookmarking synced across devices
Start Researching
Academic Attribution
Instant Multi-Format Citation Playground
Maintaining academic integrity is seamless with 1-click citation export. Select a format below to preview the formatted reference for a sample CTU capstone study.
Cadaro, J., Cabardo, C., Conde, C., De Jesus, A., & Ompoy, R. (2025). IoT-Driven Water Quality Monitoring and Automated Aeration for Coastal Aquaculture in Northern Cebu. Cebu Technological University — Daanbantayan Campus Library.
Supports direct copy & export across all project detail views in RE-CAPS.
Under the Hood
Production-Grade Modern Tech Stack
RE-CAPS is architected with a decoupled full-stack model, combining high-speed vector retrieval with server-sent event streaming and cloud storage.
Vector Database & AI Engine
Pinecone Vector DB
recaps-projects-search index (multilingual-e5-large)
4-Tier AI Engine
Mistral → Groq → Gemini → OpenRouter cascading LLMs
Backend & Cloud Database
Node.js & Express
REST API, SSE streaming, and Pinecone vector sync
Google Firebase
Firestore NoSQL, Firebase Auth (RBAC), and Hosting
Digital Assets & Communications
Cloudinary CDN
Cloud image hosting, thumbnail transforms, and lightbox
Resend & NodeMailer
Two-factor OTP email verification and feedback routing
Academic Institution
CTU Daanbantayan Campus Library
Cebu Technological University — Daanbantayan Campus
Located in Agujo, Daanbantayan, Cebu, Philippines. The campus library serves as the central scholarly repository for students, faculty, and researchers. RE-CAPS bridges physical accession records with 24/7 digital cloud intelligence, safeguarding institutional knowledge for generations to come.
Agujo, Daanbantayan, Cebu College of Technology BS in Information Technology A.Y. 2026–2027
The Engineering Team
Engineered by Group 2, BSIT 3A
College of Technology · Cebu Technological University — Daanbantayan Campus · A.Y. 2026–2027
JC
John Cadaro
Project Lead & Backend Dev
Node.js · Pinecone · REST
CC
Carmela Cabardo
AI Integration & RAG
Cascading LLMs · Embeddings
CC
Christine Conde
Information Architecture
Metadata · Requirements
AJ
Allan De Jesus Jr.
UI/UX Design & Frontend
Responsive UI · Aesthetics
RO
Rheina Ompoy
Database & QA Engineer
Firestore · Security Rules
Have feedback or found an issue?
Reach out to the RE-CAPS development team — we value your suggestions and bug reports.