Enable partial and wildcard search on column names in IBM Knowledge Catalog

Currently, IBM Knowledge Catalog allows searches on column names only when the search term matches the beginning of the column name.

For example, considering a column named IE_EMITENTE:

After opening support case TS022814916, IBM Support confirmed that this behavior is currently working as designed and that the backend search index does not support prefixed wildcard searches. Support also confirmed that no documented limitation or alternative search mechanism was identified and recommended raising an enhancement request for product management review.

Business Problem

In real-world data catalog environments, users frequently know only part of a field name and often search using a keyword that appears in the middle or at the end of a column name.

Examples:

• Searching for EMITENTE should return assets containing columns such as:

  • IE_EMITENTE

  • NM_EMITENTE

  • CD_EMITENTE

• Searching for CNPJ should return assets containing columns such as:

  • CD_CNPJ

  • NR_CNPJ

  • DS_CNPJ

Today, these assets are not returned unless the search term matches the beginning of the column name.

This significantly reduces the effectiveness of data discovery and metadata exploration, especially in large enterprise catalogs containing thousands of assets and columns.

Expected Behavior

The search engine should support one or more of the following:

• Partial matching within column names.

• Substring search.

• Wildcard search (for example: *CNPJ*, *EMITENTE*).

• Configurable "contains" search behavior.

• Indexing of full column names so searches can match any part of the field name.

Benefits

• Improved data discovery experience.

• Faster identification of relevant datasets and assets.

• Reduced effort for business users, data stewards, analysts, and data engineers.

• Better usability of IBM Knowledge Catalog in large-scale environments.

• Increased value of metadata indexing and catalog governance capabilities.

Use Case

A user knows the business attribute they need, such as:

• CNPJ

• CPF

• EMITENTE

• CLIENTE

• CONTRATO

However, the user does not know which dataset, table, or asset contains that information.

A search for the business term should return all assets containing columns that match the searched term, regardless of its position within the column name.

Support Case Reference

Related IBM Support Case: TS022814916

IBM Support confirmed the current behavior is working as designed and suggested creating an enhancement request for Product Management evaluation.