Thanks for the responses, Pei.
So what seems to be the prevailing and most mature deployment mode out there - deposit xml formatted input files in a given directory and cTakes write the results back in a similar format? Any idea how this compares in speed and scalability to the db input/db output option? Regarding JdbcCollectionReader (input) and CasConsumer (output) enabling - are there ways to add and configure those from the visual tool, CVD (as it is done in the ctakes i2b2 tool on the i2b2 wiki)? Also, where do you specify the mapping of the db columns to the internal filed names so cTakes knows what to process (like i.e. in the DataHandler's section of the solrconfig.xml file)? If somebody has implemented a similar cTakes config and would be willing to share the details and the wisdom we will appreciate tremendously. Regards, vlad From: Chen, Pei [mailto:[email protected]] Sent: Monday, September 16, 2013 6:12 PM To: [email protected]; [email protected] Subject: RE: configuring cTakes for getting input from a db table and persisting output to i2b2 and/or flat db table > to a) take input from a ms sql db table Something like a JdbcCollectionReader? http://svn.apache.org/repos/asf/ctakes/trunk/ctakes-core/src/main/java/org/a pache/ctakes/core/cr/JdbcCollectionReader.java >b) configure it so that it writes NLP results back to the db Something like a CasConsumer? http://svn.apache.org/repos/asf/ctakes/trunk/ctakes-core/src/main/java/org/a pache/ctakes/core/cc/CasConsumer.java >, in i2b2 format as well as in a flat table format. There was some early/prelim work done in sandbox on the i2b2 formats, but never really made it to mainstream ctakes (yet.). http://svn.apache.org/repos/asf/ctakes/sandbox/ctakes-gui/src/main/java/org/ chboston/cnlp/ctakes/gui/runtime/result/ --Pei From: Vlad Valtchinov [mailto:[email protected]] Sent: Monday, September 16, 2013 5:13 PM To: [email protected] Subject: configuring cTakes for getting input from a db table and persisting output to i2b2 and/or flat db table Hello cTakes Users- we're relatively new to cTakes (even though not to NLP) and are currently evaluating cTakes 3.0 for moving some of our clinical radiology DSS pipeline to Apache cTakes. We have an evaluation install under Linux, and would like to start with configuring cTakes to a) take input from a ms sql db table, and b) configure it so that it writes NLP results back to the db, in i2b2 format as well as in a flat table format. Somebody out there has probably implemented this already - can you point us to an online resource where this is detailed, or even better share binary implementation with us? cTakes is supposedly very similar to another Apache project, Solr, and yet we could not find a clear documentation of DataHandler configuration under cTakes... maybe not looking at the right place... Thanks much in advance. Vlad Valtchinov Brigham Radiology, CEBI
