cancel
Showing results for 
Search instead for 
Did you mean: 

MDG-M Score in duplicate check

RiccardoBrogi
Product and Topic Expert
Product and Topic Expert
0 Kudos

Hi guys,

I configured TREX and duplicate check on a MDG-M 6.1 system.

Enterprise Search works fine, however I faced a "unexpected" behavior in duplicate check: the system identifies as potential duplicates only records that perfectly fulfill the specified matching criteria and the returned similarity score is ALWAYS 50.00.

On MDGIMG, I tried to modify the Low and High Threshold under General Settings -> Data Quality and Search -> Search and Duplicate Check -> Configure Duplicate Check for Entity Types and the weights for every relevant field under General Settings -> Data Quality and Search -> Search and Duplicate Check -> Define Search Applications: no changes at all!

Is anyone facing this problem?

Regards,

Riccardo

Accepted Solutions (1)

Accepted Solutions (1)

RiccardoBrogi
Product and Topic Expert
Product and Topic Expert
0 Kudos

Hi,

I received the confirmation than Enterprise Search (ES mode) does not currently support weights/ranking and so the displayed similarity score is meaningless.

I hope this would be helpful for you.

Regards,

Riccardo

Former Member
0 Kudos

Hi Riccardo, Thank you! i also noticed that the score is always 50.

What do you think about fuzzy search? It seems to be a problem with it too.

Situations:

1) Create material with a template -> "Check" -> we get 50 score duplicate

2) Create material with a template -> Change 1 symbol in description -> "Check" -> No potential duplicates found (Even if material with changed description already exists)

Didnt you face that issue?

Former Member
0 Kudos

Hi Alex,

For material , only Enterprise Search (ES) can be used as the search mode for duplicate check so no fuzzy search(BAS-AD) as search mode for duplicate check for material is supported.We are currently working on HANA Search(HA) for material which will be part of MDG 7.0 SP02 release.

So as  mentioned by Riccardo Enterprise Search (ES mode) does not currently support weights/ranking, so overall  in material duplicate check ,weights/ranking is not supported.

Thanks & Regards,

Ibrahim.

Former Member
0 Kudos

Thank you, Ibrahim.

It seems that the standard ES app for MDGM (100% duplicate only) is not always enougth.

So, is there an opportunity to maintain custom fuzzy search (e.g. ABAP distance function or oracle utl_match) within the MDG 6.1?

Regards, Alex.

Girisha
Advisor
Advisor
0 Kudos

Hello Alex,

ES does the Fuzzy search with a pre-configured fuzzy value(not sure where the fuzzy value is configured, i can check that).

You can check the release note of ES to find more details.

Best regards,

Girish

Former Member
0 Kudos

Hi Girish!

I guess you are talking about Threshold values (low and high) in "Configure duplicate check for entity types" customizing, but they dont affect with ES search mode.

In debugger i see that fuzzy mode get disabled with ES as well...

So is it true that MDG-M standard duplicate check only can "catch up" 100% similar records?

Girisha
Advisor
Advisor
0 Kudos

Hello Alex,

Ya threshold's values doesn't affect ES.

If the Fuzzy is enabled in the IMG for ES, then the fuzzy should work for duplicate check.(See my attachment for enabling the fuzzy).


And for your second question the duplicate check should give you the result where the match is 70%(I think the value 0.7 is defaulted in ES) for fuzzy enabled fields.


You already know that rank is not supported in ES.


Best regards,

Girish


Former Member
0 Kudos

Hi Girish.

Yes, this fuzzy option is activated by default, but "Check" button in MDG UI only shows a duplicate when it is 100% equal

I also noticed with debugger, than fuzzy search parameter goes to "FALSE" in CL_SDQ_USMD_SEARCH_DATA_IMPL - IF_USMD_SEARCH_DATA~EXECUTE method...

Didnt found the solution yet.

Girisha
Advisor
Advisor
0 Kudos

Hello Alex,

I did a quick check in our systems and it works with fuzzy, see my attachment for more details where I've given the description as "tes", in the duplicate check popup I got the results with "tes" "Tes" and "TES KM", the records with "Tes" and "TES KM" are result of the Fuzzy search.

Best regards,

Girish

PS: Fuzzy is supported only for the operator EQ(Equals).

Former Member
0 Kudos

Girish, thank you!

Maybe something is wrong with the connector, i double checked MDG-side customizing ..

Can you please tell me where i can check matching operators?

Girisha
Advisor
Advisor
0 Kudos

Hello Alex,

Sorry i didn't get your question regarding the matching operators. The place where we set the FUZZY to false is in the EXECUTE method of the class CL_SDQ_USMD_SEARCH_DATA_IMPL.

You can check the that once in debugging, which will ensure that the issue is not the IMPL class.

You can send me the mail with your contact details so that we can have a quick chat.

Best regards,

Girish

Former Member
0 Kudos

Hello,

The note :1922709 might help resolve the issue. I suppose the fuzzy value for ES is set to 7.0 by default. It is done in method : GET_STAGING_RESULTS of class : CL_SDQ_USMD_SEARCH_DATA_IMPL when setting the search mode and I am not sure if this value can be changed by any existing configuration.

Best regards,

Shefz.

0 Kudos

Girish,

Could you please help me to configure the duplicate check for material. We are unable to get popup window.

Regards,

Lakshmi

Answers (2)

Answers (2)

RiccardoBrogi
Product and Topic Expert
Product and Topic Expert
0 Kudos

Hi Tommy,

yes, even when you make a copy of an existing material the score is always 50.

Former Member
0 Kudos

even when you copy the material to a new material and the similiarity score is 50?