Fwd: ZIP file with Tomcat and Sesame pre-configured

[Tom Munnecke]

Here is a neat file to set up a Sesame server from my buddy Mark Watson who wrote a book on the subject http://www.markwatson.com/opencontent/book_java.pdf

---------- Forwarded message ----------
From: Mark Watson <mark....@gmail.com>
Date: Thu, Dec 1, 2011 at 8:22 PM
Subject: ZIP file with Tomcat and Sesame pre-configured
To: Tom Munnecke <munn...@gmail.com>


Tom, grab the file http://markwatson.com/tomcat-sesame-rdf.zip

unzip the file and:

cd tomcat-sesame-rdf

and run:

bin/catalina.sh run

hit URL:

http://localhost:8080/openrdf-workbench

or, more interestingly:

http://localhost:8080/openrdf-workbench/repositories/actors/explore?resource=%3Chttp%3A%2F%2Fdbpedia.org%2Fresource%2FBest_Actress_Academy_Award_winners%3E

If you want to do more than just play with this, read the docs
carefully: http://www.openrdf.org/doc/sesame2/users/

Enjoy,
Mark

--
Mark Watson, consultant and author: http://markwatson.com

"No art, however minor, demands less than total dedication if you want
to excel in it." Leon Battista Alberti

[George Lilly]

Thanks Tom. This is great. What a find!

gpl

[glilly]

I installed tomcat-sesame-rdf on my laptop... then on my desktop.. and
had so much fun loading our sample rdf files and browsing through them
that I also installed it on one of our test servers. You can access it
here:

http://vistaewd.net:8980/openrdf-workbench/repositories/gpltest/contexts

I uploaded a few of the qds quality measure files (a couple of the
large ones won't load for some reason... and the full qds.rdf file
won't load for some reason either)...

I also uploaded 4 of the test smart patients..

Just click on a context and you can see what's in the rdf file... I
really want one of these rdf browsers for our triplestore... Rob, can
we do this in ewd?

Please feel free to create your own repositories on this site or just
upload additional rdf files to the gpltest one that's there...

gpl

(ps. it really ran much slower on my home machines, which are
Windows.. this linux install works nicely... gpltest has only 2
indexes, the default ones but it manages with those - it is a "native
java" repository, no MySql)

On Dec 2, 12:27 am, Tom Munnecke <munne...@gmail.com> wrote:
> Here is a neat file to set up a Sesame server from my buddy Mark Watson who

> wrote a book on the subjecthttp://www.markwatson.com/opencontent/book_java.pdf

>
>
>
>
>
>
>
> ---------- Forwarded message ----------
> From: Mark Watson <mark.wat...@gmail.com>
> Date: Thu, Dec 1, 2011 at 8:22 PM
> Subject: ZIP file with Tomcat and Sesame pre-configured

> To: Tom Munnecke <munne...@gmail.com>
>
> Tom, grab the filehttp://markwatson.com/tomcat-sesame-rdf.zip

>
> unzip the file and:
>
> cd tomcat-sesame-rdf
>
> and run:
>
> bin/catalina.sh run
>
> hit URL:
>
> http://localhost:8080/openrdf-workbench
>
> or, more interestingly:
>

> http://localhost:8080/openrdf-workbench/repositories/actors/explore?r...

[Tom Munnecke]

Yes, I got a kick out of interactively adding triples to the files.  It really drove home how things work (for me)... I can only stare at specs so long before my fingers get itchy...

I just signed up to talk about semantic stuff at the VistA community meeting in January... 

Conor and I had a meeting with James Fowler, author of Connected http://connectedthebook.com/ who has done a lot of exciting research in the area of social networks and health:

Here are some links to some of our social network papers so you can see the kind of things we do with this data.

http://jhfowler.ucsd.edu/spread_of_obesity.pdf
http://jhfowler.ucsd.edu/collective_dynamics_of_smoking.pdf
http://jhfowler.ucsd.edu/spread_of_alcohol_consumption.pdf
http://jhfowler.ucsd.edu/dynamic_spread_of_happiness.pdf
http://jhfowler.ucsd.edu/alone_in_the_crowd.pdf
http://jhfowler.ucsd.edu/social_network_sensors.pdf

Connecting clinical data to social network data is an amazing new area of medical research; and the VA has the scale to make this possible.  There are some really interesting issues about scale and privacy: if he has a large enough sample size, he can bin the data in such a way as to deidentify patients better.  Triple store is a great way to handle social network graphs, I think.  

Note that Allegro Graph has some interesting social network stuff: http://www.franz.com/agraph/allegrograph/ (I hear that their license fee can run $40K per server, by the way.)  I had a lot of trouble trying to get their Gruff visual tool to work, which is when I gave up and asked Mark for help - which led to this zip file.

[Tom Munnecke]

are you guys looking at how VistA can play in the Smart Platform?

http://wiki.chip.org/smart-project/index.php/Developers_Documentation:_SPARQL_Examples_for_SMART#Find_Medication_Names

I like what I see regarding the SPARQL semantics for the app interface...

[George Lilly]

Tom:

The short answer is yes. 

The not-quite-as-short answer (the "approach") is that we are first spending some effort to read, store, and process smart and other rdf before trying to generate it from the VistA patient record. It's possible that the fileman triple store will be an effective intermediary/enabler inside VistA for processing incoming and outgoing smart rdf - among many other equally useful applications.

It seems to me that it will be a whole lot easier for VistA to provide (and use) smart containers if it has an embedded triple store. 

gpl

[Rob Tweed]

Yes, a native triple store implementation for GT.M would be the ideal, In the meantime, Allegro Graph will provide a good way of determining what's needed and getting familiar with the technology, terminology and appreciating the potential of the semantic healthcare web.

Global storage lends itself very easily and nicely to representing triples and graphs and providing a high-performance and scalable platform - that's the easy bit.  The hard/laborious bit is the SPARQL query parser/optimiser, though of course one approach would be to emulate Allegro Graph's database access protocols (I don't know what they are or look like), in which case you could use the same tools you're currently using (though I personally have an aversion to the Java stack, Tomcat etc - way too many moving parts that you have to fiddle about with, but that's just me).

When I get the time I want to look at some of the Javascript-based stuff which seems promising,

Rob

--
Rob Tweed
Director, M/Gateway Developments Ltd
http://www.mgateway.com
------------------
EWD Mobile: Build mobile applications faster
http://www.mgateway.com/ewd.html

[LD 'Gus' Landis]

Tom/George,

You guys are way ahead of me, but I really like what I am

hearing reported.

Once things are firmly in mind (needed functionality), it would

be good to make this "triple store" stuff native to Fileman (as

there should be a "dom" file type, etc).

I know that Rick and crew identified like seven new file types

needed in Fileman for the laboratory package.

As to making things "EWD"-able... That should be rather straight

forward (Hmmm.. I see that Rob just posted something, a nice

gmail feature, so it will be interesting to see his reaction as I'll risk

it and post this before reading his response!)

Anyway it is wonderful that there are some working prototypes for

this stuff so that we can know what we want before implementing

it in the "correct technology" (MUMPS), perhaps as an option that

is indistinguishable from the current implementation (e.g. Rob's

Amazon storage option that invisibly is hosted by a GT.M instance).

Cheers,

  --ldl

--
---
NOTE: If it is important CALL ME - I may miss email,
which I do NOT normally check on weekends nor on
a regular basis during any other day.
---
LD Landis - N0YRQ - de la tierra del encanto
3960 Schooner Loop, Las Cruces, NM 88012
651-340-4007  N32 21'48.28" W106 46'5.80"

[Tom Munnecke]

You might be interested in the architecture of the Bigdata scalable store - that is based on B-tree implementation http://www.bigdata.com/whitepapers/bigdata_whitepaper_10-13-2009_public.pdf

It looks like Sesame SAIL is the common interface between triplestores:

"It should however also work with other triple-stores that have a Sesame Sail implementation (such as AllegroGraph, Neo4J, Mulgara, Owlim, and 4Store). Some Triple-stores don't offer a Sail layer, but do have a Sesame Repository interface to access the store (e.g. Virtuoso). You can still use the USeekMSail for such stores, by wrapping the store specific Repository in a RepositorySail (see uSeekM Library Documentation) first."   https://dev.opensahara.com/projects/useekm/wiki/USeekMSail

Rob,

I'm sorry I didn't get smart on EWD while in Seattle, but I think that the most exciting part of all this technology is the possibility of semantic interoperability with the rest of the world... which is what I see RDF as good for.  I am particularly interested in the Smart Platform architecture of an EHR "container" (e.g. a VistA instance) that communicates via RDF at the App level.  For example, this is the SMART Javascript to access a medications list.

SMART.api_call({
                  method: 'GET',
                  url: "/records/" + SMART.record.id + "/medications/",
                  data: {}
               },

       function(contentType, data) {
                  alert('data received: ' + data);
               });

This is all very interesting stuff to me... offering the possibility that we could deploy a "Semantic Overlay" model of interoperability, with an RDF-based "Link and Go" approach rather than go down the disastrous path of "plug and play" mega monoliths.   It would introduce a notion of an "App Store" model of lots of approved apps available for download, as well as provide a semantic basis from which to deal with privacy and security.

From what I can tell from the Smart Platform architecture, we would need to expose VistA as a "Container" 

here is an example SPARQL query to a Container:

PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>

PREFIX sp: <http://smartplatforms.org/terms#>

PREFIX dcterms: <http://purl.org/dc/terms/>

SELECT  DISTINCT ?t

  FROM <http://sandbox-api.smartplatforms.org/records/2169591/medications/>

WHERE {

  ?m rdf:type sp:Medication .

  ?m sp:drugName ?medc.

  ?medc dcterms:title ?t.

}

which looks like something that Conor's Semantic VistA already does.

The benefit of using an RDF-based approach is the ability to mashup other databases.  A disease code from a medical record could be linked to a bioinformatics database and a geospatial reasoner.  We could talk about semantic temporal reasoning systems, backward-chaining workflow models (e.g. create a "successful discharge" event and let the system pull the workflow from its desired endpoint), and all kinds of innovative stuff... 

I think that the single most important factor in getting this to work is making it simple for people to use.  There is a huge learning curve/abstraction wall associated with RDF and semantic web stuff, so the simpler we make the process - and the earlier the user gets to see a "Hello World" app, the more likely it's success.

I'd like to schedule a Google+ Hangout next week on this topic... maybe we can get Mark Watson to join us...   any takers?