This repository contains a simple, REST-based blackboard architecture implementation. This implementation provides some basic plumbing for a blackboard and example knowledge sources that interact with it. A running instance of this code is available at https://blackboard.ncats.io/.
./activator run
If all goes well, then try the following:
curl --data-ascii @asthma.json -H 'Content-Type:application/json' http://localhost:9000/blackboard/api/kg
where asthma.json is a file containing the following payload:
{
"type": "query",
"name": "A simple blackboard example",
"term": "asthma"
}
Now try
curl http://localhost:9000/blackboard
There should be a new knowledge graph created with a single query node. To update this knowledge graph using the Pharos knowledge source we can use the following call:
curl -X PUT http://localhost:9000/blackboard/api/kg/1/ks.pharos
If all goes well, now rerun
curl http://localhost:9000/blackboard/api/kg/1
to see that the knowledge graph has been updated with new nodes contributed by the Pharos knowledge source. Note that each time the Pharos knowledge source is run, it makes contributions to the knowledge graph by iterating over all the available nodes in the graph. After the first iteration, the knowledge graph contains nodes that are related to the query. Let's extend these nodes further by rerunning the knowledge source:
curl -X PUT http://localhost:9000/blackboard/api/kg/1/ks.pharos
The knowledge graph should now contains new nodes and edges. To see a
more visually pleasing knowledge graph, you can shutdown
activator by hitting Ctrl+D then point your
Neo4j client to the
database blackboard.db. Once logged in, you can issue this query
match(n:`KG:1`) return n
This should return a knowledge graph similar to the one shown here:
To expand on a specific node in the KG:
curl -X PUT http://localhost:9000/blackboard/api/kg/1/nodes/2/ks.pubmed