nav[aria-label="Primary Navigation"] { padding: 0; & ul { list-style: none; width: 100%; display: flex; flex-direction: row; justify-content: start; align-items: start; gap: 30px; padding: 0; & li { margin: 0; } & ul li { list-style: none; } } }

Connecting to Graph Database from JMP or SAS

Jordan_Hiller

I seem to get all the oddball questions.

Does anybody know about connecting to graph databases from JMP or SAS? These are not your standard SQL databases that can be accessed through ODBC: Graph database - Wikipedia, the free encyclopedia

If you know anybody on the SAS side of the house I could talk to, that would be helpful as well.

Thanks all.

Jordan

Accepted answers

stan_koprowski

Hi Jordan,

I tried a couple of options directly within JMP using the Neo4j DB successfully.

What I found is that you don't actually need to run a curl command.

If you know what command you need one could fairly quickly get the data for analysis and visualizing in JMP...

You can use Internet open against the neo4j db using various commands as follows--

This will return JSON that can then be parsed and loaded as a JMP data table.

See below for some examples--

Best,

Stan

open("http://127.0.0.1:7474/db/data/relationship/types/");

"[\!"ACTED_IN\!",\!"DIRECTED\!",\!"PRODUCED\!",\!"WROTE\!",\!"FOLLOWS\!",\!"REVIEWED\!"]"

open("http://127.0.0.1:7474/db/data/labels/");

"[ \!"Movie\!", \!"Person\!" ]"

open("http://127.0.0.1:7474/db/data/propertykeys/");

"[ \!"rating\!", \!"roles\!", \!"born\!", \!"released\!", \!"summary\!", \!"name\!", \!"tagline\!", \!"title\!" ]"

//Get the first node's properties

open("http://127.0.0.1:7474/db/data/node/1/properties");

\!"born\!" : 1964,

\!"name\!" : \!"Keanu Reeves\!"

// Get the 2nd node's properties

open("http://127.0.0.1:7474/db/data/node/2/properties");

\!"born\!" : 1967,

\!"name\!" : \!"Carrie-Anne Moss\!"

// Get the 10th node's properties

open("http://127.0.0.1:7474/db/data/node/10/properties");

\!"tagline\!" : \!"Everything that has a beginning has an end\!",

\!"released\!" : 2003,

\!"title\!" : \!"The Matrix Revolutions\!"

All comments

Chris_Kirchberg

Hi Jordan,

Without ODBC, JMP cannot connect. I am not sure about SAS, but this type of data is relationship based (ontology like data). Not sure how one would use in JMP due to its complex nature and non-data table like structure. My guess is SAS probably could (Look as SAS/ACCESS).

SAS/ACCESS Software | SAS

Chris

stan_koprowski

Hi Jordan,

If your customer is using the open source Neo4j as the graph DB it might be possible to access via curl commands that would return JSON. You could then use xan@jmp JSON parser to populate into a JMP data table.

I believe you could also do a similar pull using the R interface with JMP directly although I have not tried it myself.

Some additional details as to which graph DB and what type of queries they are running would help to understand the problem space better.

Stan

stan_koprowski

Hi Jordan,

I tried a couple of options directly within JMP using the Neo4j DB successfully.

What I found is that you don't actually need to run a curl command.

If you know what command you need one could fairly quickly get the data for analysis and visualizing in JMP...

You can use Internet open against the neo4j db using various commands as follows--

This will return JSON that can then be parsed and loaded as a JMP data table.

See below for some examples--

Best,

Stan

open("http://127.0.0.1:7474/db/data/relationship/types/");

"[\!"ACTED_IN\!",\!"DIRECTED\!",\!"PRODUCED\!",\!"WROTE\!",\!"FOLLOWS\!",\!"REVIEWED\!"]"

open("http://127.0.0.1:7474/db/data/labels/");

"[ \!"Movie\!", \!"Person\!" ]"

open("http://127.0.0.1:7474/db/data/propertykeys/");

"[ \!"rating\!", \!"roles\!", \!"born\!", \!"released\!", \!"summary\!", \!"name\!", \!"tagline\!", \!"title\!" ]"

//Get the first node's properties

open("http://127.0.0.1:7474/db/data/node/1/properties");

\!"born\!" : 1964,

\!"name\!" : \!"Keanu Reeves\!"

// Get the 2nd node's properties

open("http://127.0.0.1:7474/db/data/node/2/properties");

\!"born\!" : 1967,

\!"name\!" : \!"Carrie-Anne Moss\!"

// Get the 10th node's properties

open("http://127.0.0.1:7474/db/data/node/10/properties");

\!"tagline\!" : \!"Everything that has a beginning has an end\!",

\!"released\!" : 2003,

\!"title\!" : \!"The Matrix Revolutions\!"

MikeD_Anderson

@stan_koprowski - I just got a question about this same topic on Friday. I was wondering if this solution still holds? Would using the REST API maybe be another route? Has anyone else had requests for Graphic Databases? Sorry - lot's of questions here... I'm basically wondering what has changed about the solution in light of the new capabilities since 2015.

Thanks!

M

stan_koprowski

Hi @MikeD_Anderson

Do you know which graph database your customer is enquiring about?

As stated previously, what often is returned is JSON. So one could use directly without much overhead of writing http requests.

One could use load text file () and json to data table() to hit the endpoint.

Most of the graph databases have entries into the endpoints from javascript, .NET, python, JAVA, and probably others.

I haven't looked at neo4j in a while but they do have an HTTP API that could be used but that might be over engineering the problem.

If you could share a bit more about the problem and what they would like to do we might be able to offer better suggestions.

cheers,

Stan

note: endpoint can be thought of as a table. The table might only have a few observations or many depending on the endpoint definition.