Desktop product · v1.0.1
HDLE Premium
Desktop Hebrew lexicography & terminology extraction — the same Hebrew-NLP core as LinguistPro, built for professionals.
What it's for
HDLE Premium (Hebraic Dynamic Lexicon Engine) is a desktop workbench for professional Hebrew lexicography and terminology extraction — aimed at translators, terminologists and researchers rather than learners. Import documents (TXT, DOCX, PDF with OCR), define reference corpora, and extract the terminology that matters from your own material.
How it works
A 387k-document Hebrew Wikipedia baseline ships as a reference corpus. The engine extracts multi-word expressions and lemmatized terms with PMI / T-score scoring, backed by FTS5 concordance search and a translation memory. Batch translation runs through Google Cloud Translation API v3 (the official client) with budget guards and usage tracking, with a free fallback tier.
Built with Python + PyQt6, SQLAlchemy + SQLite (WAL, FTS5), and Stanza for Hebrew lemmatization and
POS tagging. Credentials are encrypted at rest with AES-256-GCM in the OS keyring —
nothing leaves the machine except calls to the translation provider the user explicitly selects.
Projects export to portable .hdleproj bundles (ZIP + SHA256) and to Excel, CSV, JSONL,
TBX and TMX.
Built on Google Cloud
- Cloud Translation v3
Planned / in evaluation
- Vertex AI
- Document AI
Recent changes
- Reference-corpus integration (M12) in progress; local NLLB offline translation and adaptive worker batching.
- Project Exchange — portable .hdleproj bundles with SHA256 integrity for cross-machine sharing.
- Security hardening (P0): AES-256-GCM credential encryption, OS-keyring storage, audit logging.