The paper introduces DivAlign, a four-stage pipeline for personalized research ideation that aims to reduce AI-induced homogenization without sacrificing researcher-direction fit. It extracts fine-grained researcher profiles, generates profile-conditioned candidates, scores them for Executability, Comprehensibility, and Growth Potential, and reduces redundancy across a community portfolio. On a benchmark covering 95 AI researchers across five subfields, DivAlign lowers average pairwise similarity from 0.331 to 0.294 and nearest-neighbor similarity from 0.704 to 0.608, while retaining 99.9% of the fit score relative to an independent top-choice variant.
No heat snapshots are available in the last 24 hours.