Analytics:
- Add video watch metrics section to shop dashboard (avg % played,
completion rate, seeks, rewinds, pauses, speed) with per-product
video breakdown table
- Add per-product analytics detail page at /s/{shop_id}/analytics/{product_id}
with views over time, video watch, traffic, devices, ring entries,
engagement, comment sentiment, and price history
- Product title links on shop analytics now go to per-product analytics
- Add get_sentiment_summary_for_product() helper
SPA fix:
- Update thumbnail image src and wrapping link href during DJ fade
- Clear stale extra thumbnails from previous product
- Add file_url to watch JSON for thumbnail-to-file links
589 B
589 B
Add sentiment analysis to comments
Lightweight, self-contained feature. Comments exist, sentiment scoring doesn't.
Scope
- Classify comments as positive/neutral/negative
- Use a lightweight local rule-based approach (no external API dependency)
- Store sentiment score per comment in DB
- Feed sentiment data into existing analytics dashboard (MPS-3)
Notes
- No longer blocked by engagement signal weighting — analytics is live (MPS-3)
- Products with high positive:negative ratio could surface in analytics
- Keep it simple: rule-based first, upgrade to ML later if needed