Repositories
GEUL’s public repositories live under the geul-org GitHub organization. The normative grammar is GEUL Grammar on this site.
A GEUL document is a sequence of 16-bit WORDs. The unit of data is the record, which starts with one TAG WORD, and records are held in a document. Documents are published by an Issuer, and each version is pinned by a content hash. The repositories below are tools that write, read, convert, search and count these documents, plus a Korean text analyzer.
Library
gogeul
The GEUL Go library and the geul command-line tool.
Writes, reads, checks and publishes GEUL documents (.geul). Its packages provide format checks, record decoding through Codec spec documents, version diffs, IMPORT dependency resolution and the TOULMIN Codec. The geul CLI covers editing (new, add), publishing (publish), browsing (ls, log, show, diff, check), a browser viewer (serve, build) and pack archives (pack build, pack unpack). Pack archives can store records compressed with zstd.
| GitHub | geul-org/gogeul |
| Language | Go |
| License | CC0 |
Wikidata
wikigeul
Wikidata → GEUL conversion (formerly gogeulwiki).
Reads Wikidata JSON dumps directly and produces one GEUL document per item or property, with labels, descriptions, aliases, statements with their qualifiers and references, and sitelinks in every language. All documents are published under the Issuer wikidata.geul.org (0x00000001).
| GitHub | geul-org/wikigeul |
| Language | Go |
| License | CC0 |
wikipg
Lossless loading of Wikidata JSON dumps into PostgreSQL. Deprecated since 2026-11-03; use wikigeul, which reads the dumps directly.
Stores entities, labels, descriptions, aliases, statements, qualifiers, references, sitelinks and Lexemes in dump order, and keeps every dump line verbatim. The dump itself (file, URL, size, checksums, loader commit) is recorded too. wikipg does not depend on GEUL.
| GitHub | geul-org/wikipgeul |
| Language | Go |
| License | CC0 |
Search
silk
SILK — an indexing and search engine built for GEUL.
Builds partitioned, sorted indexes of the records in a GEUL pack archive and answers queries such as “which documents state that someone’s occupation (P106) is physicist (Q169470)?” without decoding every document. It frames records and their fields from TAG WORDs alone, without a codebook. It is in early development, validating its search algorithms, and its formats may change before a first release.
| GitHub | geul-org/silk |
| Language | Go |
| License | CC0 |
Statistics
statistics
Counts TAG usage in GEUL pack archives, per Issuer.
Counts records for each 9-bit code left after dropping the Length from a TAG WORD (TAG IDs 0x000L–0x1FFL). Given several archive folders or pack set folders, it counts them together.
| GitHub | geul-org/statistics |
| Language | Go |
| License | CC0 |
Korean
koparse
A Korean text analyzer.
It starts by separating nouns from the particles and endings attached to them in Korean words (eojeol), and will grow into a morphological analyzer if needed. It is built from scratch rather than on existing Korean morphological analyzers or dictionaries, and its dictionary uses only CC0 sources (Wikidata labels and Lexemes) and self-generated data. It is in early development.
| GitHub | geul-org/koparse |
| Language | Go |
| License | CC0 |
Codebooks (pre-v1)
The two repositories below were designed before the current v1 grammar. Parts of them do not match it.
sidx
Research on SIDX (Semantic-aligned Index) assignment.
Research material toward assigning SIDX, a semantic address that gives close numbers to entities with close meanings, from Wikidata usage statistics. It holds the earlier 48-bit Entity attribute schema, encoder scripts and codebooks; no SIDX assignment has been produced yet. The normative definition of SIDX is GEUL-SIDX in the grammar.
| GitHub | geul-org/sidx |
| Language | Python |
| License | CC0 |
wordnet-verb
A provisional 16-bit verb codebook based on WordNet 3.0.
Gives each of the 13,767 verb synsets of WordNet 3.0 one 16-bit code, with close codes for verbs close in meaning. It is a provisional edition to be replaced by a codebook that does not use WordNet, and how the codes would be carried in GEUL records is not decided.
It contains material derived from WordNet, so it is not a CC0 output. The WordNet-derived parts follow the WordNet 3.0 License; only the GEUL project’s own parts, such as the codes and the classification, are CC0.
| GitHub | geul-org/wordnet-verb |
| Language | Python |
| License | WordNet 3.0 License (derived parts) · CC0 (own parts) |
Website
geul-org
Source of this website, geul.org.
A Hugo static site in 12 languages, deployed with S3 + CloudFront.
| GitHub | geul-org/geul-org |
| Language | Hugo |
| License | CC0 |