Triple Store/Graph Database Overview There have been discussions in the past between Mayo and NCI...

Preview:

Citation preview

Triple Store/Graph Database

• Overview• There have been discussions in the past

between Mayo and NCI about using a triple store/graph database. The Semantics Infrastructure (SI) group of CBIIT has also done considerable research and analysis on this topic.

• We need to understand what NCI’s requirements are surrounding this topic.

Triple Store/Graph Database

• Discussion topics to understand NCI’s requirements:

• Is there an existing requirement to move to a triple store or graph db?

• Does the SI group have recommendations? If not, when will they have them?

• Will the EVS team be following the recommendations of the SI group?

Triple Store/Graph Data Base

• SI Talking Points • Are federated queries required?

• Should these be over SPARQL endpoints? • There are potential security issues with

this.• Does SPARQL need to support

reasoning/inference?• What priority is performance?

• Why is performance only focused on for triple stores?

• How does this tie in to any LexEVS/CTS2 API requirements, if any?

Triple Store/Graph Data Base

• What does a terminology datastore need to do?

• KVS - Key-Value store

• DS - Document Store

• RDBMS - Relation Database

• GDB - Graph Database

Datastore Feature Datastore Type that Performs Well

Store a resource with an ID KVS, DS, RDBMS, GDB

Find a resource by ID KVS, DS, RDBMS, GDB

Find a resource by a set of properties DS, RDBMS, GDB

Find all edges of a resource GDB, RDBMS

Traverse a graph GDB

Compute subgraphs GDB

Perform set operations on subgraphs GDB

Calculate paths GDB

Triple Store/Graph Data Base

• Demo of ArangoDB

• Next steps/roadmap

Triple Store/Graph Data Base

• Discussion Recap• Requirements • Decision Points • Priority

Recommended