Introduction to Modeling

The basic premise of Modeling in Custodian is the separation of matching and the storage of tables of data thus ensuring that when data tables change, are added or deleted the match rules are not impacted. This obviously means that there would be some extra investment in time into the mapping of the two sides of the data model.

Domain: is a subject area of data that represents a key part of your business, such as product, customer, reference, etc. Domains contain all the objects listed below:

Concept Groups: are groupings of inter-related Concepts, they are used as a convenience for defining Indexes and Match Rules, to remove the need to define a match rule for each table of data in the system.

Concepts: is a breakdown of the Concept Group into its parts, such as an address concept group being defined as the address lines (concept) and the postcode (concept). Each of these sub levels are defined to be able to matched together.

Index: is a mechanism to bucket together similar records, this is often referred to as grouping, pigeon-holing, or clustering, essentially we are using a small section of columns in the data to create a series of keys to group the records into smaller buckets that we can then use the match rules to compare within. One index is defined as one concept group, but multiple indexes can be defined.

Match rules: Enable us to define what makes good for match, where some records have been bucketed together we need to use a match rule to be able to score the similarity between them. Match rules can use one or more concept groups and multiple match rule scan be defined.

Table: A metadata grouping that represents a set of columns, Relationship Table: is the same but can only be used to store data to further describe a relationship.

Column: Physical column of data, a collection of columns of data (Structure) or a list of values.

Map: An ordered mapping between the physical column of data in a table and the concept in the concept group.

Foreign Key: A link from one column in a table to another table of data in the same or different domain.