IBM WebSphere Information Analyzer and Data Quality Assessment

Chapter 2. Financial services business scenario 427

2.4.2 Step 2: Identify differences between the sources and targets

Differences between the source and target can be determined as follows:

򐂰 From active relational database catalogs, dictionaries, repositories, and other

documentation that provide details about the metadata of the various data

sources and targets.

More often than not, metadata information stored in sources (other than the

active relational database catalogs) tends to be out of date because it is not

maintained regularly as systems evolve.

򐂰 From an analysis of the data itself using tools such as IBM WebSphere

Information Analyzer. Metadata is deduced from the data and presented to

the data analyst for review and affirmation or denial of the deduction.

These are complementary approaches, essential to achieving a fuller

understanding of how synchronized the definition of the metadata is with the data

content. It also enables you to keep the metadata about data sources current,

which is critical for building new systems that require data integration from

existing systems.

Data elements in the target not found in the source,

and target data element not nullable or has no

default values

򐂰 Define default values in the target; some

impact on target applications

Code mismatch between the source and the target

data elements; for example, a salutation can be Mr,

Mrs, Dr, Miss, and Ms, and source and target do not

have corresponding codes

򐂰 Coarse to fine

򐂰 Fine to coarse

򐂰 Coarse to fine

– Use transformation to perform the mapping

򐂰 Fine to coarse

– Transform with loss of granularity; loss of

function

– Modify target definition to match fine

granularity of the source; significant impact

on target applications likely

Multiple data elements in the source maps to a

single data element in the target

򐂰 For example, an address

򐂰 Use transformation to perform the mapping

Single data element in the source maps to multiple

data elements in the target

򐂰 For example, an address

򐂰 Use transformation to perform the mapping

– Might require standardization software

Different character maps such as Unicode and

ASCII

Transformation to perform the mapping

Commonly encountered differences Potential actions