metadata
title: OpenAIRE Graph Evaluator Documentation
emoji: 📚
colorFrom: blue
colorTo: indigo
sdk: static
pinned: false
license: cc-by-4.0
short_description: Research paper classifier and LLM-as-Judge pipeline
OpenAIRE Graph Evaluator Documentation
Rendered documentation for the Echo-DSRN research-intent evaluator — a multi-model LLM-as-Judge pipeline built for the OpenAIRE AI Hackathon 2026.
This Space is a static build of the project's MkDocs site. Source lives in the
OpenAIRE-AI-Research-Evaluator
repository under docs/.
Contents
- Quick Start — run the classifier and the collaborative hub
- Architecture — Echo-DSRN intent classifier + multi-model LLM-as-Judge pipeline
- LLM-as-Judge — the judge protocol, verdict schema, and dataset export
- MCP — agent-facing classification via Server-Sent Events
- OAuth — HuggingFace login for the collaborative hub
- Inference — local and remote inference
- Hackathon Story — question, journey, insight, reusability
- Deployment — Docker Compose, Helm/k3s, PostgreSQL
Datasets & models
- Research-Intent-Judge — per-model LLM-as-Judge exports
- Research-Intent-Collab — consensus golden dataset
- Echo-DSRN-v0.1.3-Research-Intent-CLF — the classifier
License
CC-BY 4.0.