This commit is contained in:
2026-08-16 23:07:32 +02:00
parent 389608ea56
commit 6e77154228
14 changed files with 1189 additions and 21 deletions
+3 -2
View File
@@ -9,7 +9,7 @@
## Pipeline (canonical `full`)
```
normalize -> parse_specs -> fill_fields -> eprel -> ai_enhance
normalize -> parse_specs -> fill_fields -> eprel -> categorize -> ai_enhance
```
| Step | Role |
@@ -18,7 +18,8 @@ normalize -> parse_specs -> fill_fields -> eprel -> ai_enhance
| `parse_specs` | Parse `specifications` into attributes |
| `fill_fields` | Fill missing scalars / standard fields |
| `eprel` | Fetch EU energy label when `eprel_id` present + enricher on |
| `ai_enhance` | Green Chat rewrite of name + description |
| `categorize` | When mapped/prior category is empty: Pinecone vector (if configured), else LLM picks a **company taxonomy `unique_id`** from Available Categories (never invents). Persists to `processed_products.category` and empty `mapped_data.category`. |
| `ai_enhance` | Green Chat rewrite of name + description (uses category formulas/overlays when set) |
## Prerequisites