MUALLIMA: Multidomain Unified Academic Large-scale Language Infrastructure for Modeling Azerbaijani (v1.0)

## Why MUALLIMA exists Of the 4,194 Azerbaijani-affiliated records in Crossref, **none** carries license metadata. Only **two** Azerbaijani journals are indexed in DOAJ. Existing Azerbaijani corpora are dominated by social media and news text with no usable license information. MUALLIMA addresses this gap: every document in the corpus has a **verified, evidence-backed license record**. ## Corpus summary (v1.0) | Property | Value | |---|---| | Documents | 3,320 peer-reviewed articles | | Words | ~8.5 million | | Time span | 2018–2026 | | Languages | Azerbaijani 71%, English 22%, Russian 6% | | Sources | Azerbaijani academic journals (whitelisted, DOI-registered) | | Release scope | **Verified pool only** (see licensing model) | | Format | JSON Lines (one document per line) |