inno-idea-eval

Multi-persona idea evaluation uses three independent reviewer personas to assess research ideas across five dimensions. It generates per-dimension scores, a novelty grounding report, and an area-chair meta-review to produce an aggregated decision and actionable feedback.

An area-chair meta-review synthesizes independent critiques from multiple reviewer personas into a single coherent evaluation. It integrates per-dimension scores and novelty verification to produce a final aggregated decision and refinement guidance for the research idea.

Yes, evidence blocks serve as inputs alongside the idea text and pipeline state to score research ideas. The evaluation workflow processes these inputs to generate per-dimension scores, a novelty grounding report, and actionable refinement guidance.

The best way to verify novelty is through an integrated evaluation pipeline featuring a Senior ML Researcher, Domain Expert, and Methods Specialist. This multi-persona workflow generates a novelty grounding report and an area-chair synthesis with actionable feedback.

The idea evaluation workflow requires the idea text, evidence blocks, and the current pipeline state as inputs. Processing these elements allows the workflow to generate per-dimension scores, a final decision, and refined guidance.

Using multiple reviewer personas ensures rigorous critique from diverse expert viewpoints, such as a Domain Expert and a Methods Specialist. This multi-perspective approach provides a comprehensive quality gate before the area-chair meta-review synthesizes the final decision.