mrleo1nidandClaude Opus 5.5
3afed02f23
ci / recommender / recommender (push) Successful in 37s
- Implicit scores from play events and reactions per the signals table (90-day half-life, weaker skips outside the wave, dislikes override). - numpy ALS for implicit feedback; time-split evaluation on the last 7 days (NDCG@10, Recall@50) against a popularity baseline; a new model is activated only if NDCG@10 does not regress by more than 2%. - Vectors are written to track_embeddings.cf and user_profiles_rec.taste_cf, versions to model_versions; NightlyTrainer runs at TRAIN_HOUR and on first start, `pipu-rec train` runs it manually. - Wave: cf and item_cf candidate sources and a 0.10 cf term in the v1 score; impressions are tagged heuristic-v1+cf. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
95 lines
3.2 KiB
Python
95 lines
3.2 KiB
Python
"""Сервис «Моей волны»: собирает пачку из каталога, профиля, сессии и контекста."""
|
|
|
|
from __future__ import annotations
|
|
|
|
from typing import Literal
|
|
|
|
import numpy as np
|
|
import psycopg
|
|
from pydantic import BaseModel, Field
|
|
|
|
from pipu_rec.wave.contexts import Contexts, Rule
|
|
from pipu_rec.wave.engine import Pick, build_batch, model_version, taste_centroids, update_session
|
|
from pipu_rec.wave.store import IndexHolder, SessionStore, load_profile, seed_tracks
|
|
|
|
MAX_SERVED_MEMORY = 300
|
|
|
|
Outcome = Literal["complete", "skip", "skip_early", "like", "dislike"]
|
|
|
|
|
|
class Feedback(BaseModel):
|
|
track_id: str
|
|
outcome: Outcome
|
|
|
|
|
|
class Seed(BaseModel):
|
|
type: Literal["track", "album", "artist"]
|
|
id: str
|
|
|
|
|
|
class BatchRequest(BaseModel):
|
|
user_id: str
|
|
session_id: str
|
|
activity: str | None = None
|
|
mood: str | None = None
|
|
seed: Seed | None = None
|
|
exclude_ids: list[str] = Field(default_factory=list, max_length=1000)
|
|
feedback: list[Feedback] = Field(default_factory=list, max_length=200)
|
|
n: int = Field(default=5, ge=1, le=20)
|
|
|
|
|
|
class BatchTrack(BaseModel):
|
|
id: str
|
|
source: str
|
|
score: float
|
|
|
|
|
|
class BatchResponse(BaseModel):
|
|
tracks: list[BatchTrack]
|
|
model_ver: str
|
|
|
|
|
|
class WaveService:
|
|
def __init__(self, dsn: str, redis_url: str, contexts: Contexts) -> None:
|
|
self.dsn = dsn
|
|
self.index = IndexHolder(dsn)
|
|
self.sessions = SessionStore(redis_url)
|
|
self.contexts = contexts
|
|
|
|
def rules(self, activity: str | None, mood: str | None) -> list[Rule]:
|
|
"""Занятие и настроение комбинируются — пересечение профилей; неизвестный ключ игнорируется."""
|
|
rules: list[Rule] = []
|
|
if activity and activity in self.contexts.activities:
|
|
rules += self.contexts.activities[activity].rules
|
|
if mood and mood in self.contexts.moods:
|
|
rules += self.contexts.moods[mood].rules
|
|
return rules
|
|
|
|
def batch(self, req: BatchRequest) -> BatchResponse:
|
|
index = self.index.get()
|
|
with psycopg.connect(self.dsn) as conn:
|
|
profile = load_profile(conn, req.user_id)
|
|
seeds = seed_tracks(conn, req.seed.type, req.seed.id) if req.seed else None
|
|
session = self.sessions.get(req.session_id)
|
|
update_session(
|
|
index, session, [(f.track_id, f.outcome) for f in req.feedback], taste_centroids(index, profile)
|
|
)
|
|
picks: list[Pick] = build_batch(
|
|
index,
|
|
profile,
|
|
session,
|
|
self.rules(req.activity, req.mood),
|
|
n=req.n,
|
|
exclude=set(req.exclude_ids),
|
|
seed_ids=seeds,
|
|
rng=np.random.default_rng(),
|
|
)
|
|
session.served = (session.served + [p.id for p in picks])[-MAX_SERVED_MEMORY:]
|
|
artists = [index.artist[index.pos[p.id]] for p in picks]
|
|
session.recent_artists = (session.recent_artists + artists)[-10:]
|
|
self.sessions.put(req.session_id, session)
|
|
return BatchResponse(
|
|
tracks=[BatchTrack(id=p.id, source=p.source, score=p.score) for p in picks],
|
|
model_ver=model_version(index, profile),
|
|
)
|