Skip to main content
The features described in this section are currently in beta testing and are subject to change.
The feature is enabled on a per-project basis. Contact your Cognite account representative to enable the feature for your projects.
The smarter search results feature enables additional parsing and tokenization to improve searches for tag-like naming structures. Tag-like structures usually consist of letters and numbers, optionally delimited by various symbols, such as PX-14-A-0059 or AB42/9x14.

​
Affected fields

When you enable smarter search results for a project, the additional tokenization rules apply to the following fields:
  • CogniteDescribable.name
  • CogniteDescribable.description
  • CogniteDescribable.tags
  • CogniteDescribable.aliases

​
Tokenization rules

For fields affected by smarter search results, the following tokenization rules apply in addition to the standard tokenization rules:
  • All symbols split tokens. A symbol is any character that is not a letter or a number. For example, AB-42 splits into two tokens: AB and 42. Similarly, AB/42 splits into AB and 42.
  • Number-letter transitions split tokens. For example, AB42 splits into two tokens: AB and 42. Similarly, 42AB splits into 42 and AB.
  • Leading zeroes are removed from tokens. For example, AB0042 splits into two tokens: AB and 42. This lets you search for ab42 and find the tag without having to remember how many leading zeroes the tag has.
  • Camel case transitions split tokens. For example, AbcDef splits into two tokens: Abc and Def.
Last modified on February 17, 2026