feat(reco): cold-start, decay, MMR, anti-genre, reasons, eval + full … - #75
Merged
Conversation
…signal coverage Implements the 10 proposed algorithm improvements and ensures every tracked profile field actually influences scoring. New module server/utils/recoScoring.ts (scoreCandidate + applyMMR): - calibrated 0-100 match score with labeled contributions - localized 'why' reasons surfaced on MediaCard (fr/en/de reco.reason.*) - negative-keyword / anti-genre / anti-actor / anti-director demotion - genre x moment + genre x month (seasonality) terms - sessions-to-finish effort signal; household/co-viewing kids sub-profile blend - deriveUserStyle() consumes 100% of profileData fields (global biases) - MMR diversity re-ranking (genre/collection similarity) youMightAlsoLike rewrite (media.ts): - cold-start trending/popular fallback rail - epsilon-greedy exploration (~15% of slots, outside top genres) - MMR re-rank; reasons + 0-100 matchScore exposed - reads per-title stats from cached materialized snapshot (off hot path) analyticsEngine.ts: - applyRecencyDecay() — ~75d half-life decay of all taste maps (decayAt) - genreMonth seasonality recording; household kids/adult sub-profiles + co-viewing mediaStatsEngine.ts: getAllMediaStatsCached() materialized 60s cache. recoEval.ts + analytics.evaluateReco (admin): offline precision@k / recall@k / NDCG@k / MRR harness for objective weight tuning. Docs updated (recommendation-algorithm.md, WHATS_WATCHED.md). https://claude.ai/code/session_01RF3jY6P1ySdMQzuv97NFtL
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
…signal coverage
Implements the 10 proposed algorithm improvements and ensures every tracked profile field actually influences scoring.
New module server/utils/recoScoring.ts (scoreCandidate + applyMMR):
youMightAlsoLike rewrite (media.ts):
analyticsEngine.ts:
mediaStatsEngine.ts: getAllMediaStatsCached() materialized 60s cache.
recoEval.ts + analytics.evaluateReco (admin): offline precision@k / recall@k / NDCG@k / MRR harness for objective weight tuning.
Docs updated (recommendation-algorithm.md, WHATS_WATCHED.md).
https://claude.ai/code/session_01RF3jY6P1ySdMQzuv97NFtL