Hi All, I had the same problem. So I used TopBraid Composer to make a SPARQL motion script that takes the XML in and converts it to RDF and used SPARQL to get to the triples that I was interested in. But this required manually reading the XML to see how things were organized.
I tried the RDF Consumer, but wasn't able to get it to work very well. So I went with what I knew. Has that project progressed any since mid-summer? LFJ On Mon, May 6, 2013 at 2:53 PM, <"Austin Ziehlke"@nike.apache.org> wrote: > Thanks**** > > Is there a workflow for these tools that I have not seen? Like use CVD to > parse a Medical Record, Then run the results XML through CPE to produce RDF > like info?**** > > Just looking to get as much info as possible.**** > > ** ** > > *From:* Murali Nagendranath [mailto:[email protected]] > *Sent:* Monday, May 06, 2013 2:50 PM > *To:* [email protected] > *Cc:* [email protected] > *Subject:* Re: Managing cTakes application output**** > > ** ** > > Austin,**** > > ** ** > > You may want to check out an existing RDFConsumer class: > http://svn.apache.org/repos/asf/uima/sandbox/trunk/RDFCASConsumer/src/main/java/org/apache/uima/rdf/RDFCASConsumer.java > **** > > ** ** > > This might shed additional insight into the output format and schema you > are trying to get to.**** > > ** ** > > Murali**** > > ** ** > > On Mon, May 6, 2013 at 2:41 PM, <"Austin Ziehlke"@athena.apache.org> > wrote:**** > > Pei**** > > Thanks, this is helpful. I will look for the UIMA CAS Viewer and see what > that brings. I am still investigating options, and understanding how to get > info from medical record into a triple store is my ultimate goal. But > cTakes puts out way too much info. Before I invest in custom software > solution I want to make sure I am looking at a good path.**** > > **** > > Any further info you might have will be a blessing to me.**** > > **** > > Best Regards**** > > Austin**** > > **** > > **** > > *From:* Chen, Pei [mailto:[email protected]] > *Sent:* Monday, May 06, 2013 11:28 AM > *To:* [email protected] > *Subject:* RE: Managing cTakes application output**** > > **** > > Hi A.Ziehlke,**** > > Note: The CVD (CAS Visual Debugger) is an UIMA tool which is really > helpful visualizing a single result mainly for debugging purposes. There > are many ways to customize the output of cTAKES. Were you just looking for > something simple to visualize the xml output? If so, then there is a tool > called UIMA CAS viewer [1] where you can import the xml and pick and choose > the Types you are interested in.**** > > Alternatively, there are programmatic ways to customize the output: Some > examples include:**** > > 1) Writing a custom Java CASConsumer [2] where you can select the > Types that your application may be interested in.**** > > 2) There is a GU[3] I in the sandbox area which is intended to make > this easier for users to configure.**** > > **** > > Hope that helps.**** > > [1] > http://uima.apache.org/d/uimaj-2.4.0/tutorials_and_users_guides.html#ugr.tug.cpe.configurator_and_viewer > **** > > [2] > http://svn.apache.org/repos/asf/ctakes/trunk/ctakes-core/src/main/java/org/apache/ctakes/core/cc/ > **** > > [3] http://svn.apache.org/repos/asf/ctakes/sandbox/ctakes-gui/**** > > **** > > *From:* AZiehlke @ knology [mailto:[email protected]<[email protected]>] > *On Behalf Of *"Austin Ziehlke"@nike.apache.org > *Sent:* Monday, May 06, 2013 11:15 AM > *To:* [email protected] > *Subject:* Managing cTakes application output**** > > **** > > Greetings all.**** > > **** > > I am relatively new to cTakes, and am seeking some guidance on best > use/practices. Specifically on the output generated by cTakes CVD. The XML > output is verbose. I need to make that output more succinct to user needs. > Is there a way in CVD to scale down the output, customize the output? Is > the generated output based/controlled on the data in the dictionaries? If > CVD does not allow customizing the output, do you have any suggestions on > secondary tools that will make it easier to get to the medical info and > hide/remove the lexical/grammar info. Is custom software necessary to parse > the data?**** > > **** > > When I import the xml generated from CVD directly into Allegrograph, and > display in Gruff, the resulting graph is enormous for one medical record > entry. Perhaps there is some step between the output from CVD and the input > to a datastore that can reduce the data to a manageable state?**** > > **** > > A.Ziehlke**** > > **** > > ** ** >
