;;; elfeed-score.el --- Scoring rules for elfeed -*- lexical-binding: t; -*-
;;
;; Scoring rules for ranking papers by research interest.
;; Higher scores = more relevant papers shown first.
;;
;; Research Areas:
;;   1. LLMs & AI (PRIMARY) - training, alignment, advances
;;   2. Marine Biology - cetaceans, orcas, migration, acoustics
;;   3. Pedagogy - STEAM and mathematics education

((version 8)

 ;; ============================================================================
 ;; TITLE SCORING
 ;; ============================================================================
 ("title"

  ;; ----------------------------------------
  ;; LLMs & AI (PRIMARY FOCUS) - High scores
  ;; ----------------------------------------
  ;; Core LLM topics
  (("large language model" 300)
   ("LLM" 300)
   ("GPT" 250)
   ("transformer" 250)
   ("attention mechanism" 200)
   ("BERT" 150)
   ("language model" 200)
   ;; Training methods
   ("fine-tuning" 250)
   ("fine tuning" 250)
   ("pretraining" 200)
   ("pre-training" 200)
   ("training dynamics" 200)
   ("scaling law" 250)
   ("emergent" 200)
   ("in-context learning" 250)
   ("few-shot" 200)
   ("zero-shot" 200)
   ("prompt" 180)
   ("instruction tuning" 250)
   ("RLHF" 300)
   ("reinforcement learning from human feedback" 300)
   ;; Alignment & Safety
   ("alignment" 300)
   ("AI safety" 300)
   ("AI alignment" 350)
   ("constitutional AI" 300)
   ("harmlessness" 250)
   ("helpfulness" 200)
   ("jailbreak" 200)
   ("red teaming" 200)
   ("interpretability" 250)
   ("mechanistic interpretability" 300)
   ("explainability" 200)
   ;; Architecture advances
   ("mixture of experts" 200)
   ("MoE" 200)
   ("state space model" 200)
   ("Mamba" 250)
   ("retrieval augmented" 200)
   ("RAG" 200)
   ("chain of thought" 250)
   ("reasoning" 200)
   ("multimodal" 180)

   ;; ----------------------------------------
   ;; Marine Biology - Cetaceans & Acoustics
   ;; ----------------------------------------
   ("cetacean" 300)
   ("whale" 250)
   ("orca" 350)
   ("killer whale" 350)
   ("dolphin" 250)
   ("porpoise" 200)
   ("marine mammal" 250)
   ("baleen" 200)
   ("odontocete" 250)
   ("mysticete" 200)
   ;; Intelligence & Behavior
   ("whale intelligence" 350)
   ("cetacean cognition" 350)
   ("dolphin cognition" 300)
   ("echolocation" 300)
   ("bioacoustics" 300)
   ("marine acoustics" 300)
   ("whale song" 300)
   ("vocalization" 250)
   ("click train" 250)
   ("acoustic" 150)
   ;; Migration & Ecology
   ("migration pattern" 300)
   ("migratory" 200)
   ("pod" 200)
   ("whale population" 250)
   ("marine conservation" 200)
   ("ocean" 100)

   ;; ----------------------------------------
   ;; Pedagogy - STEAM & Math Education
   ;; ----------------------------------------
   ("mathematics education" 300)
   ("math education" 300)
   ("STEM education" 250)
   ("STEAM" 300)
   ("pedagogy" 250)
   ("teaching" 150)
   ("learning" 100)
   ("student" 100)
   ("classroom" 150)
   ("curriculum" 200)
   ("educational" 150)
   ;; Math-specific
   ("mathematical thinking" 300)
   ("problem solving" 200)
   ("proof" 150)
   ("mathematical reasoning" 250)
   ("numeracy" 200)
   ("algebra education" 250)
   ("geometry education" 250)
   ("calculus education" 250)
   ;; Methods
   ("active learning" 200)
   ("inquiry-based" 250)
   ("constructivism" 200)
   ("Montessori" 200)
   ("project-based" 200)

   ;; ----------------------------------------
   ;; Deprioritize
   ;; ----------------------------------------
   ("survey" -30)
   ("review article" -20)
   ;; Hide political/society topics
   ("politics" -500)
   ("political" -500)
   ("policy" -200)
   ("social media" -300)
   ("misinformation" -300)
   ("disinformation" -300)
   ("fake news" -400)
   ("election" -500)
   ("democracy" -400)
   ("governance" -300)
   ("regulation" -200)
   ("ethics" -100)
   ("bias" -50)
   ("fairness" -50)))

 ;; ============================================================================
 ;; AUTHOR SCORING - Researchers to follow
 ;; ============================================================================
 ("authors"
  (;; LLM researchers
   ("Ilya Sutskever" 400)
   ("Andrej Karpathy" 350)
   ("Yoshua Bengio" 350)
   ("Yann LeCun" 350)
   ("Geoffrey Hinton" 350)
   ("Dario Amodei" 400)
   ("Chris Olah" 350)
   ("Jan Leike" 350)
   ;; Marine biology (add your followed researchers)
   ;; Pedagogy (add your followed researchers)
   ))

 ;; ============================================================================
 ;; FEED SCORING - Prioritize by source
 ;; ============================================================================
 ("feed"
  (;; LLM feeds get boost (primary focus)
   ("cs.LG" 100)
   ("cs.CL" 100)
   ("cs.AI" 80)
   ;; Marine biology
   ("biorxiv" 50)
   ("plos" 50)
   ("nature" 60)
   ;; Pedagogy
   ("physics.ed-ph" 40)
   ("math.HO" 40)))

 ;; ============================================================================
 ;; CONTENT/ABSTRACT SCORING
 ;; ============================================================================
 ("content"
  (;; LLM keywords in abstract
   ("state-of-the-art" 50)
   ("SOTA" 50)
   ("benchmark" 30)
   ("OpenAI" 50)
   ("Anthropic" 80)
   ("DeepMind" 50)
   ;; Marine biology
   ("Pacific Northwest" 100)
   ("Salish Sea" 100)
   ("Southern Resident" 150)
   ;; General
   ("novel" 30)
   ("breakthrough" 40))))