Under Settings → System → Document Classification, you define rules that automatically check incoming documents and tag them with a hashtag (tag). When a rule’s content condition matches a document, the document receives the configured tag. This tag is the engine behind automation – it lets you:
- Trigger workflows automatically (via a start rule),
- automatically attach classified documents as an enclosure,
- or feed the documents in specifically as LLM context (e.g. via
TAG(...)in the PromptToLLM task).
A classification rule
Clicking + Regel hinzufügen opens the rule editor. When it matches, a rule assigns one or more hashtags and consists of conditions, which you combine using AND/OR groups.
| Component | Meaning |
|---|---|
| Name / Description | Label and purpose of the rule. |
| Hashtags | the tags assigned when the rule matches. |
| Rule type | the kind of match – e.g. whether a specific phrase appears in the document. |
| Phrase | the text to search for. |
| Concatenation | combines multiple conditions into groups with AND / OR, so even complex patterns can be modeled. |
| Status | Enabled or Disabled; rules can also be imported and exported. |
Statistical vs. AI-based classification
This page describes statistical (rule-based) classification – it checks fixed content criteria (rule type + phrase). Alternatively, DEPLAW can classify documents via a language model (workflow task ClassifyCaseFiles).
| Approach | When it makes sense |
|---|---|
| Statistical (rule-based, this page) | for documents with a recurring structure – court filings, invoices, and similar. Recommended for its high performance, stability, and low resource use. |
| AI-based (LLM) | for content that varies or unstructured documents where fixed phrases aren’t enough. More flexible, but more compute-intensive. |