The oko reasoning engine is performing forward and backward chaining of n3p data.
The oko engine is processing n3p data and it is itself implemented as n3p data.
It is running on top of Scryer Prolog which is implemented in Rust.
| TERM | Examples |
|---|---|
| IRI | 'http://example.org/ns#Socrates' |
| VARIABLE | X _abc |
| LITERAL | "abc" 1.52 1e-18 pi dt("2022-01-15",'http://www.w3.org/2001/XMLSchema#date') |
| LIST | [TERM,...] [TERM,...|LIST] [] |
| TRIPLE | IRI(TERM,TERM) |
| GRAPH | TRIPLE,... |
| CLAUSE | Examples |
|---|---|
| ASSERTION | TRIPLE. true => GRAPH. |
| FORWARD_RULE | GRAPH => GRAPH. |
| QUERY | GRAPH => true. |
| ANSWER | GRAPH => true. |
| INFERENCE_FUSE | GRAPH => false. |
| BACKWARD_RULE | TRIPLE <= GRAPH. |
The oko engine performs forward chaining for a FORWARD_RULE and backward chaining for a BACKWARD_RULE.
Queries are posed and answered as GRAPH => true. so the answers are also queries be it with
some parts substituted and eventually containing more variables than in the original query.
This forms a dialogue leading to necessary and sufficient answers, supported by proof steps, so that action can take place.
The builtin triples are:
'http://idlabresearch.github.io/ns#builtin'(Subject,[])'http://idlabresearch.github.io/ns#pso_triple'([P,S,O],Triple)
Install Rust and Scryer Prolog and then
$ git clone https://github.com/idlabresearch/oko
$ cd oko
$ ./test
- Personal notes by Tim Berners-Lee: Design Issues
- Book of Markus Triska: The Power of Prolog