gitea-admin/gh-christianlouis-docuelevate
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

docs: add classification rules API and user guide documentation

Browse Source 
Co-authored-by: christianlouis <361235+christianlouis@users.noreply.github.com>
This commit is contained in:
copilot-swe-agent[bot]
2026-03-09 23:39:09 +00:00
parent df051e8b81
commit bd898605de
2 changed files with 176 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files
docs/UserGuide.md
+19 -1
View File
@@ -573,7 +573,25 @@ Processing pipelines let you define exactly what happens to your documents when
| `embed_metadata` | Write extracted metadata into the PDF document properties |
| `compute_embedding` | Compute semantic embeddings for similarity search |
| `send_to_destinations` | Upload the processed document to all configured storage destinations |
| `classify` | Classify the document type with AI |
| `classify` | Classify the document type using rules (filename patterns, content keywords, metadata) |
#### Classify step – rule-based document classification
The `classify` step assigns a category to each document by evaluating **built-in** and **custom** classification rules. Rules are matched against three signals:
- **Filename patterns** — regex matched against the original filename (e.g. `(?i)invoice` matches filenames containing "invoice").
- **Content keywords** — pipe-separated keywords matched against the OCR text (e.g. `invoice number|amount due`).
- **Metadata match** — `field=value` matched against existing AI metadata (e.g. `document_type=Invoice`).
**Pre-built categories** include: Invoice, Contract, Receipt, Letter, Report, Bank Statement, Tax Document, Insurance, and Payslip. You can also define your own custom categories.
The classification result is stored in the document's `ai_metadata` under the `classification` key with the matched category, confidence score, and list of matched rules. If no `document_type` was previously set by AI metadata extraction, the classify step will also populate it.
> **Tip:** Manage custom classification rules via **Settings → Classification Rules** or the `/api/classification-rules/` API. See the [API Documentation](./API.md#classification-rules) for details.
| Option | Type | Default | Description |
|--------|------|---------|-------------|
| `use_builtin_rules` | boolean | `true` | Include the pre-built classification rules |
#### OCR step options