domain-language-extraction

Apply the five extraction techniques: noun phrase extraction for domain nouns, entity pattern recognition for business objects and events, acronym detection, relationship extraction, and action extraction for domain verbs. Results are organized into a structured ubiquitous language dictionary.

A ubiquitous language dictionary is a shared vocabulary of domain terms agreed upon by developers and business stakeholders. It contains entities, personas, domain events, processes, business rules, metrics, and acronyms with precise definitions used consistently across code and documentation.

Confidence is frequency-based: 1.0 for terms appearing more than 10 times consistently, 0.9 for 5–10 occurrences, 0.8 for 2–4 occurrences, and below 0.8 for ambiguous terms. Low-confidence terms are flagged as needing clarification.

Yes, extracted terms integrate directly into domain-model.md. Terms not found in the existing model are flagged as uncertainty factors, helping teams identify gaps between documentation and the current domain model.

Terms appearing fewer than two times receive low confidence scores and may be missed or misclassified. Ambiguous terms with context-dependent meanings require human clarification, and vague terms are listed as terms to avoid rather than resolved automatically.