Captions, descriptions, linguistic metadata

Three families of translatable strings live in a Power BI semantic model. Daxlate’s grid edits the first two directly; the third is preserved byte-for-byte on round-trip.

Captions #

User-visible names of tables, columns, measures, hierarchies, and hierarchy levels. This is what a report consumer sees in the field list, on visuals, and in tooltips. Captions are the primary thing translators work on.

Descriptions #

Free-form metadata, surfaced in tooltips and AI-driven explorers (Q&A, Copilot). Often skipped in translation workflows; the v1 grid surfaces them as first-class so you can decide per-culture whether to translate or leave at the source.

linguisticMetadata #

The Q&A engine reads a JSON blob attached to the model that maps natural-language phrasings to model objects (“show me revenue by region” → the Revenue measure and the Region column). Editing this blob is out of scope for the v1 grid; Daxlate reads it, preserves it on round-trip, and never silently drops it. A dedicated editor is on the long-term roadmap.