Documentation Center

Implementing Semantic AI

Implementing Semantic AI consists of implementing its various subfeatures in different parts of the product architecture. The current topic gives a broad overview of a Semantic AI implementation. Semantic AI is a non-core feature; implementing it is not required. Similarly, while some subfeatures depend on others, not all subfeatures necessarily need to be implemented.

About this task

In the list below:
  • Steps 1 and 2 are mandatory; steps 3, 4 and 5 are optional.
  • Steps 1 and 2 must be performed first; steps 3, 4 and 5 can be done later, and in parallel.

The implementation plan below is a suggestion; you can use your external taxonomy for other purposes too, such as designing an intelligent, taxonomy-based website navigation, for example.

Procedure

  1. Use Taxonomy Space to create an external taxonomy for your content.
    What this means: Your organization obtains predefined taxonomies for your business domain, either from the public domain or commercially, and import it into the taxonomy provider. Content designers can then customize and modify the taxonomy you obtained to define your business domain in a controlled vocabulary.
  2. Connect to the external taxonomy from the Tridion Sites product core.
    What this means: An application administrator adds the Tridion Taxonomy Connector to the product and configures it, to allow both Content Manager and Content Delivery to connect to, and interact with, the external taxonomy. More information at Connecting to external taxonomies in Taxonomy Space.
  3. Automatically tag your content with concepts from the external taxonomy using smart tagging.
    What this means: Authors and editors can enrich their content with metadata, which is automatically generated using the smart tagging feature in the Experience Space user interface. Depending on the strictness of the regulatory requirements in your industry, authors and editors may subsequently still need to modify the set of automatically generated tags. More information at Classifying content with taxonomies.
  4. Implement a faceted search interface based on the external taxonomy.
    What this means: Web developers can use the Public Content API to construct GraphQL queries that both retrieve concepts (search facets) and their possible values to show to the website visitor, and execute the faceted search queries that those visitors create. More information at GraphQL samples for faceted search and in the documentation of the GraphQL API itself.
  5. Implement a search suggestion feature based on the external taxonomy.
    What this means: Web developers can use the Public Content API to construct GraphQL queries that match an entered search term to relevant content through concept matching. Such queries maximize search efficacy by finding synonyms or closely related concepts of the entered search term. For example, a user who enters the term "notebook" would yield results for content tagged with "laptop" or even "computer" (a related concept). More information at GraphQL sample for search suggestions and in the documentation of the GraphQL API itself.