How to configure OpenEMPI to use probablistic match?

Hello, everyone. I am working on Registration module enhancement project. You can discover what we have done on my midterm video:

We have chose OpenEMPI as MPI solution for our project, but we have some problems with it now. We can’t configure OpenEMPI to use “Probablistic match” instead of “Exact match”. There seems to be a configuration in deep of project, but we can’t find it and documentation of OpenEMPI doesn’t has any turtorial about this.

So, if you have experience with OpenEMPI - it would be just great, if you can assist us with configuring Probablistic match :smile:

1 Like

While you may find some people here on OpenMRS Talk who know the answer to your OpenEMPI answer, have you checked the OpenEMPI forums for answers?

The spin I would put on this is:

  • the purpose of this GSoC project is to work on the OpenMRS side of integration with MPI, not for @approce to learn how to configure OpenEMPI
  • I asked him not to spend significant time on this, since its exact matching is sufficient for our GSoC purposes
  • The idea is that any implementation actually using OpenEMPI will put in the time to configure it right.
  • however this will be a much better proof-of-concept and demo if someone experienced with OpenEMPI can give us some quick advice about the minimal thing we’d have to do to enable probabilistic matching on the OpenEMPI server we already have set up.
  • @rcrichton maybe?

In your openempi/conf directory there should be a config file named mpi-config-probabilistic-matching.xml. Modify that to your needs and rename it to mpi-config.xml, after backing up your original mpi-config.xml of course :smile:. That should get you started along with this wiki page:

Having said that I would recommend using exact matching if that fits your needs, as I have had some trouble with probabilistic matching previously.

1 Like

Hi, just to say that I was just about to post the exact same link Ryan Yates posted. Sorry for being late to the party :slight_smile:

1 Like