3 min read

Configure the Semantics microservice

The Semantics section allows you to configure the settings for the Semantics microservice and its capabilities provided by the Semantics API and the Semantic-Similarity batch process such as Semantics recommendations carousels and catalogue enrichment, respectively.

Configure Semantics in Instance Management

warning

Only Playboard users with permissions for Instance Management can set up and edit the Semantics settings.

When you access the Semantics configuration form for the first time, the Deactivated label appears beside the Semantics section in the side menu. This means that there is no active configuration for the service. Once you have created the configuration, you need to activate the Semantics settings.

To set up the Semantics microservice settings, complete these steps:

  1. Configure the settings for the Semantics microservice.
  2. Save the configuration and activate Semantics.

Configuring Semantics microservice settings

Configure the following fields to ensure the good performance of the Semantics microservice. You don't need to configure all fields as every instance has different Semantics capabilities enabled.

note

There are fields whose configuration affects the semantic catalogue enrichment and others that affect the Semantics recommendations carousels. Field descriptions indicate each case.

  • Semantic model: Encoder model used to produce semantic embeddings. Field for catalogue enrichment only.
    The naming convention to use in this field is: <name_file>.tar.gz,<instance>,<language>,sbert (for example, semantic_model.tar.gz,spainstore,spanish,sbert).
  • Max query candidates: Maximum number of semantic suggestions to be returned per query. Field for catalogue enrichment only.
  • Max distance: Maximum distance by which the semantic suggestions returned can differ from the input query. Field for Semantics recommendations only.
  • Catalogue enrichment fields: Fields considered to find semantically related suggestions. Field for catalogue enrichment only.
  • Index token: Token to be used to authenticate against the Index microservice. Field for catalogue enrichment only.
  • Pivot fields: Product catalogue fields to use as pivots for filtering the query candidates to be returned. Field for Semantics recommendations only.
  • Valid languages: List of languages supported by the semantic model used to generate Semantics recommendations. Instance's languages not included in the list will be excluded. Field for Semantics recommendations only.
  • Min query count: Minimum number of occurrences that a query from organic interactions should have to be considered a semantic candidate. Queries with fewer occurrences than this threshold are discarded as they are too rare. Field for catalogue enrichment only.
  • Max query results count: Maximum number of results that a query from organic interactions should have to be considered a semantic candidate. Queries with more results than this threshold are discarded as they're too generic. Field for catalogue enrichment only.
  • Max Semantics candidates: Maximum number of candidates to be returned per query. Field for Semantics recommendations only.
  • Filters: List of Semantics API default filters to refine the result set of suggestions. Find more details about the supported filters on the Semantics API page. Field for Semantics recommendations only.

note

For semantic catalogue enrichment, make sure that you configure the values for the fields Semantic model, Max query candidates, Catalogue enrichment fields, Index token, Min query count, and Max query results count.

note

For Semantics recommendations, make sure that you configure the values for the fields Max distance, Pivot fields, Valid languages, Max Semantics candidates, and Filters.

Activating Semantics

Once you have updated the configuration and clicked Save, you need to activate the Semantics microservice by turning On the toggle switch.