Managing relation indexes in Data Marketplace

The Data Marketplace relation-based search and relation-based custom filters help to optimize the search experience. They allow data consumers to, for example, find data sets by entering business terms.

The relation-based search and relation-based filters feature in Data Marketplace needs to be enabled, and needs indexed relations, meaning that the relations are stored for search purposes. Relations are indexed based on defined relation indexes. Several out-of-the-box relation indexes are available. However, Data Marketplace administrators can create, update, and delete relation indexes based on the organization's needs.

  • The relation-based search automatically takes all relation indexes into account.
  • To use a relation index as a relation-based custom filter, the administrator must activate the filter in the Data Marketplace Filters settings.

Tip To watch a video to understand relation indexes better, go to Example: Create a new relation index for a custom filter.

Elements in a relation index

A relation index consists of a name and one or more relation paths.

  • The name of the filter can become visible in the Filters pane.
    • The name of a relation-based filter cannot be changed.
    • A _ (underscore) or a space in the name will translate in a space in the Filters pane.
  • A relation path represents the chain of relations that allows Data Marketplace to find assets that have a direct or indirect relation with each other.
    • A relation path starts from the asset type you want in the search results (root asset type) and ends with the asset type or attribute type you want to filter on.
    • The relation path also takes the asset type children’s relations into account.
    • A relation path can contain 1 to 4 relations.
    • A relation index can contain multiple relation paths. However, each relation path must end with the same asset type.

Questions to construct a relation index

Use the following questions to prepare for the creation of relation indexes.

Question Answer and consequence Example
Which asset type (and children) or asset types do you want to filter out?
  • If you identified one asset type, then this asset type is your root asset type, meaning the first asset type in your relation path.
  • If you identified multiple asset types, you need to create a relation path for each asset type.
Table

Which asset type or attribute type do you want to filter on? If it is an attribute type, which asset type does it belong to?

Note Only Text, Selection, Multiple Selection, Numbers, and Boolean (True/False) attribute types can be selected as attributes in a relation path.

  • This asset type is the last asset type in your relation path.
  • The asset type or attribute type should be reflected in the filter name.
Data Domain.
In your metamodel, which path do you need to follow to get from the first to last asset type in the relation-based filter? This is the relation path you will need to use. To go from Table to Data Domain, we need the following path: Table contains Column, Column is represented by Data Attribute, Data Attribute is part of Data Entity, Data Entity to Data Domain.
Which filter name is meaningful to consumers of the data?   Data Domain

Out-of-the-box relation indexes

Filter Relation paths
Data Domain Relation path for Data Domain - Data Set:
  1. Data Set contains Column
  2. Column is represented by Data Attribute
  3. Data Attribute is part of Data Entity
  4. Data Entity is classified by Data Domain
Relation path for Data Domain - Table:
  1. Table contains Column
  2. Column is represented by Data Attribute
  3. Data Attribute is part of Data Entity
  4. Data Entity is classified by Data Domain
Data Concept Relation path for Data Concept - Data Set:
  1. Data Set contains Column
  2. Column is represented by Data Attribute
  3. Data Attribute is classified by Data Concept
Relation path for Data Concept - Table:
  1. Table contains Column
  2. Column is represented by Data Attribute
  3. Data Attribute is classified by Data Concept
Business Asset Relation path for Business Asset - Data Set:
  • Data Set is related to Business Asset
Relation path for Business Asset - Report:
  • Report is related to Business Asset
Relation path for Business Asset - Table:
  • Table is related to Business Asset
Relation path for Business Asset - Table:
  1. Table contains Column
  2. Column is represented by Business Asset
Source System Relation path for Source System - Data Set:
  • Data Set is implemented into Source System
Relation path for Source System - Table:
  1. Table is part of Schema
  2. Schema belongs to Technology Asset
  3. Technology Asset is grouped by Source System
Data Source Relation path for Data Source - Table:
  • Table is part of Schema
    The data source is an attribute of the Schema asset.
Important The out-of-the-box relation index Data Set - Column, which is specific to relation-based search, is not visible and cannot be updated or removed.