Beruflich Dokumente
Kultur Dokumente
Contributing Speaker(s)
Christian Behre
NetWeaver Product Management, SAP AG
Remo Durante
Solution Architect, SAP Deutschland AG & Co.KG
Learning Objectives
INFORMATION INTEGRATION
heterogeneous IT landscape
Life Cycle Mgmt
Knowledge Mgmt
Leverages existing IT investments in business-critical data Delivers vastly reduced TCO by effective master data management ensuring cross-system data consistency Accelerates and improves the
PROCESS INTEGRATION
Integration Broker
APPLICATION PLATFORM
J2EE
ABAP
DB and OS Abstraction
Master Data Management Master Data Management PROCESS INTEGRATION Integration Broker Bus. Process Mgmt
DB and OS Abstraction
R/3
SRM
Siebel
i2
Legacy
...
II 1
BI
Staging
4 4
4 1
4 Knowledge Mgmt.
Exchange Infrastructure
3 3
CRM
Legacy
3rd Party
Legacy
Loading Master Data with Extraction MDM triggers the load PULL mechanism
II
Loading Master Data with Periodic Inbound Collector Client triggers the load PUSH principle
? =
BI
Staging
1b 1a 2
?
4
Database
7 Knowledge Mgmt.
=
Exchange Infrastructure
CRM
Legacy
3rd Party
Legacy
? =
BI
Brown & Partner Inc. 248 Meadow Lane Drive Sacramento Ca. 95816 Brownes & Partner Brown Inc.
Exchange Infrastructure
MEADOW Lane
San Diego
CA 93860
Knowledge Mgmt.
CRM
Legacy
3rd Party
Legacy
Demo
SAP AG 2004, SAP TechEd / Session MDM202 / 9
MDM cleansing capabilities provide: Normalization capabilities to store unaligned data in a comparable format Object type related Matching algorithms to compare data sets Ranking on matching results to calculate matching scores Lower and Upper Score Thresholds to pre-decide on cleansing cases
BI (BW 3.5) EP 6.0 Master Data Engine 3.00 Master Data Clients*
*i.e. SAP ERP, CRM, SRM
XI 3.00
technical routing structure mapping key-mapping cache (new 3.0)
master data administration process control (process chains) Inbound staging UIs logical routing
Upload
Validation
Matching
GetMatchingCandidates CalculateScore
Cleansing
ID-Mapping
CI
1. Upload Validation
GetNormalizedAttributeKeys
reads
ID-Mapping Table
API
writes
Mapping
reads
Score Table
writes
CalculateScore
Comparison
writes MDS
Generation of MDS-Objects
the Normalization is triggered again for the MDS Object (stops after writing into the Index Table)
Score Table
MDC-Data that lies above the lower threshold with the related calculated score mapped MDC-Objects
ID-Mapping Table
CI
1. Upload
GetNormalizedAttributeKeys
reads
NormalizeObjects
Global Normalizer
Makes the data comparable Perticular settings: Ingnore blanks, special characters or several punctuation marks Upper and lower case
Normalizing Algorithm
GetNormailzedAttributeKeys Finds the indices that are defined within the coding NormalizeObject Indicates the data and writes them into the index table
SAP AG 2004, SAP TechEd / Session MDM202 / 18
Object Type
Upper or lower case Include or exclude the characters Insert the character you want to include or exclude. Enter the Source Path / Target Path Enter the Reference Path
SAP AG 2004, SAP TechEd / Session MDM202 / 19
Global Normalizer Object model diagram of the Content Integrator Business Partner
Reference Path
Normalizing Algorithm
GetNormalizedAttributeKeys
The indices can be defined within the java coding You can define as an index
one array several arrays a certain part of an array (e.g. the first 8 characters)
Due to the fact that you can manipulate directly in the coding, the specification of the indices is very flexible.
Normalizing Algorithm
NormalizeObjects
The data are indicated and written into the index table, dependent from the previous coding settings:
Mandant Object ID consecutive numbering of data records having the same ObjectID Matched Entity ObjectID Column Key/Index Column Value
one object
NameNormalizer Customizing
At the SAP Enterprise Portal a Normalizing Algorithm Customizing in XMLFormat can be uploaded.
It may contain: The definition of NonNameTokens The substitution of several characters or a special character string The adding/truncation of several characters or a special character string
Example Normalization
Activity Example
"Mnchen Traffic Corp." 1 Remove fill signs like "# _ - / \ . ," "Mnchen Traffic Corp" 2 Convert to upper case "MNCHEN TRAFFIC CORP" 3 Normalize special characters according to predefined list (replace by AE) MUENCHEN TRAFFIC CORP 4 (Tokenize) Cut name into tokens using list of token separator: <blank>, "-" MUENCHEN, TRAFFIC, CORP 5 Check and mark token against predefined list of nonname tokens (like CORP, BANK). MUENCHEN, TRAFFIC, CORP 6 Check for minimum Length. MUENCHEN, TRAFFIC, CORP
Score Table
CalculateScore
Matching Algorithm
CalculateScore
Calculates the score between the indicated data records with the help of the settings in the XML-File
Matching Algorithm contains: the definition of the score for the particular attributes conditions for the matching check (e.g. check in a certain order with a special dependence)
The score values in a existing XML-File can be replaced without changing the Matching Algorithm.
SAP AG 2004, SAP TechEd / Session MDM202 / 29
Comment
Scoring +25 +10 +10 for first for each additional for each match of NameToken match of NameToken match of NonNameToken
Partner/SecondaryIDs/PartnerIdentification
if secID and secIDType match else +0 for each additional match of secID / secIDType pairs if secID does not, but secIDType matches if secID and taxType match else +0 for each additional match of secID / taxType pairs if secID no match, but taxType matches if no match, if match, if match, if match, for first for each additional for each for each if sum 90 if sum 90 if sum > 190 else +0 else +0 else +0 else +0 match of NameToken else match of NameToken match of HouseNumber match of NonNameToken
Partner/SecondaryIDs/TaxNumber
Score
Original Data Name Street City Postal Code DUNS Normalized Date nameTok en nameTok en non-nameTok en street street number city Postal Code DUNS Matching Result Calculated Score Final Score Result
Reference Object Mnchen Trafic Corporation Brezelstrasse 7 Mnchen 80331 4711 Normalized Values MUENCHEN TRAFIC CORP Brezelstrasse 7 MUENCHEN 80331 4711
Normalized Values Score Normalized Values Score Normalized Values Score MUENCHEN 25 MUENCHEN 25 MUENCHEN 25 RUECK 0 BROETCHEN 0 TRAFIC 10 AG 0 CORP 10 CORP 10 Brezelstrasse 15 Brezelstrasse 15 Brezelgasse 0 17 0 17 0 17 0 MUENCHEN 10 MUENCHEN 10 MUENCHEN 10 61525 0 61525 0 80331 15 4512 -10 0 4711 50
60 60 Automatic Match
120 91
4. Built decision matrix - verify with business using real life examples Name Street House No. Zip Code City Account Group Duplicate? No No? Yes? Yes
Possible Solution Adjust limit values for generating a duplicate proposal Make Wal-Mart an insignificant name part Adjust the matching strategy: Are there other identifying attributes or keys that can be taken into account?
2.
3.
2.
3.
Summary Content Integrator (CI) is a component of SAP Master Data Management (MDM) to consolidate business data in a heterogeneous system landscape. The matching process is the core process to identify identical or similar business data objects. Matching context provides the logic and conditions for matching processes. With MDM some Matching Strategies are delivered. Matching Strategies can be configured and/or developed.
Further Information
Public Web:
www.sap.com SAP Developer Network: www.sdn.sap.com Master Data Management SAP Customer Services Network: www.sap.com/services/
Questions?
Q&A
SAP AG 2004, SAP TechEd / Session MDM202 / 43
Feedback
Please complete your session evaluation. Be courteous deposit your trash, and do not take the handouts for the following session.
Thank You !
Matching Strategies
Matching Algorithm (com.sap.ci.strategies)
Mat_300_GTIN_A Mat_300_GTIN_SPN_MPN_A
Please note that this document is subject to change and may be changed by SAP at any time without notice. The document is not intended to be binding upon SAP to any particular course of business, product strategy and/or development.
SAP AG 2004, SAP TechEd / Session MDM202 / 46
To compare the two data, mark them and click the button <Compare Objects>
To merge the two objects: Set one of the data <Duplicate>, the other <Target> Set the status from <New> to <Released> Click the button <Save>