Establishing identity on the web is no easy task. The general steps for achieving this are as follows:
- Build the data warehouse for storing identities and relations. (Done)
- Seed the database with existing linked data. (In progress)
- Expose the Services in a browser. (In progress)
- Expose the Services as an API. (Not Started)
- Finalize the Specification for linking to Open GUID. (In progress)
Seed the database with existing linked data
Open GUID was initially seeded with WordNet 3.0. After review of various datasources and ontologies such as OpenCyc, YAGO, and DBPedia, and UMBEL, it was determined that WordNet was the best balance between real world entities and linkability. OpenCyc and YAGO have WordNet links already, and the others can be associated via those two. It was determined that WordNet synsets were a bit more user friendly than some of the abstract concepts found in Cyc, which was designed to be an inferencing engine.
Open GUID maintains Relations via URIs. WordNet associations are preserved by means of a fictitious URI at princeton, from which the original sysnet id can be gleaned. It is with these id’s that other links will be established. Still todo:
OpenCyc. Need to obtain the OpenCyc to WordNet synset mapping in a consumable form, and load the OpenCyc URI’s into the Relation table.
YAGO. Need to obtain the YAGO to WordNet mapping in a consumable form, and load the YAGO URI’s into the Relation table.
UMBEL. Need to obtain the UMBEL to OpenCyc URI mapping in a consumable form, and load the UMBEL URI’s into the Relation table. This includes Concepts and Named Entities. It may be a good idea to load the UMBEL named entities not in Open GUID as well, since they were created from the more complete named entity database in OpenCyc. Any duplicates would have to be merged.
DBPedia. Need to obtain the DBPedia to YAGO mapping in a consumable form, and load the DBPedia URI’s into the Relation table.
Etc. Add any more good linked data sets.
Expose the Services in a browser
Currently the browsing of Open GUIDs is enabled. The listing of keywords, description for each GUID, and tags were generated from the semantic link references in WordNet. The following needs to be done:
Edtiability. The list of keywords, GUID description, and tags need to be editable by the web community.
URI References. Still needed is the ability to modify the list of identity references, so the community can add additional ontologies and identical endpoints.
Editorial control. Similar to wikipedia, the default editability will be by everyone. However, all changes need to be reviewable and possibly voted on.
Merging GUIDs. A web interface to approve identical GUIDs and merge them is required.
Expose the Services as an API
All the basic services need to be available via an API. This includes:
Finding a list of possible GUIDs via text lookup.
Associating an existing GUID to an Open GUID.
Fetching a list of URI References.
Finalize the Specification for linking to Open GUID
Confirm the RDF statements establishing an identical relationship.
Confirm the means of embedding references (RDFa) to the GUID in content.
Confirm the association to existing ontologies.
The final timeline has not yet been established. Please consider joining the effort and Contributing to the Semantic Web.