Modeling the claims store using Neo4j

index.txt

= Modeling the claims store using Neo4j

First, let's add some items to the store.  One of the DOIs has two ORCIDs in its metadata.
//setup
[source,cypher]
----
CREATE (doi1:output {name:'10.123/abcd'})
CREATE (orcid1:person {name:'0000-1111-1111-1111'})
CREATE (doi2:output {name:'10.123/efgh'})
CREATE (doi1)-[:CreatedBy]->(orcid1)
CREATE (doi2)-[:CreatedBy]->(orcid1)
RETURN *
----
//graph

Now, we're going to import an ORCID record.  It contains one of the DOIs.
Note how the source of the information can be modeled using the direction of the relationship.
//setup
[source,cypher]
----
MATCH (doi:output{name:"10.123/abcd"})
CREATE (orcid2:person {name:'0000-2222-2222-2222'})-[:Created]->(doi)
RETURN *
----
//graph

We can now do things like find out all person-doi relationships, no matter who asserted them.

//setup
[source,cypher]
----
MATCH (works:output) -[]- (people:person)
RETURN works.name,people.name
----
//table

We can also model things like isPartOF, contains, or both!  Things with biderectional relationships are confirmed by Both parties.


//setup
[source,cypher]
----
MATCH (doi:output{name:"10.123/abcd"})
MATCH (doi2:output{name:"10.123/efgh"})
CREATE (doi)-[:contains]->(doi2)
CREATE (doi2)-[:isPartOf]->(doi)
RETURN *
----
//graph

We can add organisations, and model the relationships to suit ORCID or DOIs, as funders or creators or affiliations
//setup
[source,cypher]
----
MATCH (doi:output{name:"10.123/abcd"})
MATCH (orcid:person{name:"0000-2222-2222-2222"})
CREATE (o:org {name:'orgID'})-[:Created]->(doi)
CREATE (o)-[:Manages]->(g:grant{name:"grantID"})-[:Funded]->(doi)
CREATE (orcid)-[:Afiliated]->(o)
CREATE (g)-[:funded]->(orcid)
RETURN *
----
//graph

We can even find paths from works to funding
//setup
[source,cypher]
----
MATCH (g:grant{name:'grantID'}), (o:output{name:'10.123/efgh'}),
path = shortestPath( (g)<-[*..5]->(o) )
RETURN path
----
//table