Data sources
The open datasets behind CoolPan’s dictionaries and verb conjugation tables: their authors, licences and the changes we made.
Hebrew verb conjugation data
Hebrew verb conjugation data is based on the Hebrew Verb Lists, created by Eran Tomer as part of the master’s thesis “Automatic Hebrew Text Vocalization” at Ben-Gurion University, under the supervision of Prof. Michael Elhadad.
- Copyright:
- © Eran Tomer
- License:
- CC BY 4.0
- Source:
- github.com/NNLP-IL/Hebrew-Resources
- Data modified:
- imported into a database, PRESENT-tense forms deduplicated by grammatical person, linked to lexemes.
English vocabulary and word senses
English lexicon data (lemmas, parts of speech, word senses, and their definitions) is based on the Open English WordNet 2025 edition, a community-maintained fork and continuation of the Princeton WordNet lexical database.
- License:
- CC BY 4.0
- Source:
- github.com/globalwordnet/english-wordnet
- Attribution:
- This resource is shared under CC BY 4.0 with attribution to both Princeton WordNet and the Open English Wordnet team.
- Data modified:
- imported into a database; senses linked to lexemes and ranked by frequency (see Princeton WordNet entry below).
CEFR difficulty levels (English)
English words are tagged with a CEFR difficulty level (A1-C2) using the Words-CEFR-Dataset, which maps English words (by part of speech) to their proficiency level.
- License:
- MIT
- Source:
- github.com/Maximax67/Words-CEFR-Dataset
- Data modified:
- joined to lexicon entries by lemma and part of speech.
English sense-frequency ranking
When an English word has multiple senses, they are ranked by how often each sense actually occurs in real usage, using semantic-concordance tag counts from Princeton WordNet 3.0.
- License:
- Princeton WordNet License
- Source:
- wordnet.princeton.edu
- Data modified:
- sense tag counts joined to Open English WordNet senses by sense key; used to compute a sense_rank field.
Word frequency data
How common a word is (used to gauge how confidently a word can be resolved from context) is computed offline using the wordfreq library, which aggregates frequency data from large text corpora across many languages, including Hebrew and English.
- License:
- Apache-2.0
- Source:
- github.com/rspeer/wordfreq
- Data modified:
- frequency values precomputed offline and stored on each lexicon entry.