Topic
5 replies Latest Post - ‏2014-07-02T05:21:12Z by geets
SiddharthArun
SiddharthArun
3 Posts
ACCEPTED ANSWER

Pinned topic MDM11 - Standardizarion and Matching

‏2014-04-23T04:38:19Z |

Hi,

Let me explain my issue with a scenario, and will brief the issue later

There are different sources with different org names (but indicating the same org) about a particular organization.

Source 1 - IBM

Source 2 - International Business Machines

Source 3 - Big Blue

The requirement here is that MDM should be able to identify that all the data are the same and merge all three to a single party, and have a standardized name at the end (The standardization requirement is not yet clear, as of now we are concentrating on matching the different org names). The expectation is that the client has multiple sets of these organizations, and would like to maintain 1 unique party for each organization.

Is there a way to configure the Probability Matching Engine to look for data like this and do the matching process (any particular place where we could go and mention that IBM, International Business Machines, Big Blue all mean the same)? Or, can we use something like QualityStage to configure and perform such matching. If so, please elaborate. Any thoughts welcome. Let me know if any further information is required for understanding the issue.

Note : We are currently using MDM11 AE for this analysis.

Thanks,

Siddharth

  • RajataK
    RajataK
    4 Posts
    ACCEPTED ANSWER

    Re: MDM11 - Standardizarion and Matching

    ‏2014-05-12T12:28:37Z  in response to SiddharthArun

    Hi Siddharth,

    Even I am trying to do the same using MDMV11 . If you have found any solution on how to do that, Kindly update the solution in this forum .

    Thanks,

    Rajata

     

     

    • SiddharthArun
      SiddharthArun
      3 Posts
      ACCEPTED ANSWER

      Re: MDM11 - Standardizarion and Matching

      ‏2014-05-12T20:42:45Z  in response to RajataK

      Hi Rajata,

      Will let you know if I can configure it.

      As of now, what we have researched is that the PME is capable of such tasks (based on a configuration called nickname for Person parties and cnickname for Org parties). You can change the PME configurations using the Initaite workbench (add/update/remove nicknames, anonymous values, modify weights, thresholds), but we are still uncertain on how to go about in doing this as we are new to Initiate. If there is anyone here with expertise on Initiate Workbench, please help us with this.

      Thanks,

      Siddharth

      • MikeBryan
        MikeBryan
        1 Post
        ACCEPTED ANSWER

        Re: MDM11 - Standardizarion and Matching

        ‏2014-06-25T21:40:02Z  in response to SiddharthArun

        Hi Siddharth,

        In the Configuration for the algorithm for Standard Edition (formerly Initiate) and the pME, you can define 'Strings' that are used for a number of tasks when comparing both incoming records to match and link and search terms in searches.  One type of String can be 'Equivalences', which in normal speak are 'nicknames'.  These can be assigned to Equivalence type comparison functions, such as 'Names', where you can reasonably expect, or see, that records have names like 'IBM', or 'International Business Machines', or 'Big Blue' and you know that they all are equivalent.  You define these in the Equivalence String 'nickname', which is then used at the comparison stage of the algo and results in the name IBM and International Business Machines being compared as an equivalent.

        Cheers,

        Mike

        Updated on 2014-06-25T21:40:19Z at 2014-06-25T21:40:19Z by MikeBryan
        • SiddharthArun
          SiddharthArun
          3 Posts
          ACCEPTED ANSWER

          Re: MDM11 - Standardizarion and Matching

          ‏2014-06-26T20:33:58Z  in response to MikeBryan

          Hi Mike,

          Thank you for the response. I did try adding the values to the cnickname (for org names), and exported the project as a PME Config jar file, and added it to the MDM ear. But, when I added two separate organizations say IBM, and International Business Machines with all the other data (address, contact number, tax id) same, it still did not identify these as suspects. This cnickname is used in the algorithm for name comparison, so, it should be fine but it is not working as expected. Is there anything else that needs to be done before exporting the PME configuration? It is not urgent, we no longer need that as a requirement, just want to know for informational purpose.

          Thanks,

          Siddharth

  • geets
    geets
    2 Posts
    ACCEPTED ANSWER

    Re: MDM11 - Standardizarion and Matching

    ‏2014-07-02T05:21:12Z  in response to SiddharthArun

    Hi,

    MDM V11 is based on Osgi framework. So you would actually need to deploy the pme CBA onto WAS where MDM EBA is deployed.

     

    Thanks,

    Geetha