A system to explore, profile, and test "research taste" using LLMs.

Research taste — the implicit preferences, instincts, and aesthetic sensibilities that guide how researchers choose problems, frame questions, and evaluate work — is currently tacit and hard to articulate. This project aims to:

1. Profile research tastes from an individual's body of work, reading habits, or explicit preferences
2. Formalize taste into structured "research taste tables" that capture dimensions like problem framing style, methodological preferences, epistemic values, and aesthetic sensibilities
3. Test — run retrospective tests (does this taste profile predict past choices?) and proactive tests (what questions would this taste generate?)
4. Discover — find papers based on taste alignment (how questions are asked) rather than just keyword/topic matching

• Formal framework for "research taste" as a concept
• LLM-based taste profiling system
• Taste-based paper recommendation (structural/aesthetic matching vs keyword)
• Cultural analysis of research taste across communities
• Tools for self-reflection and taste exploration

🔬 Literature review in progress