Create and manage an Automatic Relevance Tuning (ART) model

This is for:

Developer System Administrator

Coveo Machine Learning (Coveo ML) Automatic Relevance Tuning (ART) models improve the relevance of search results.

ART learns from search and click events. If you’re using an ART model in a Coveo for Commerce implementation, it can also learn from commerce-specific user actions such as cart and purchase events. However, even in a Coveo for Commerce implementation, ART models can’t learn from commerce events alone.

To take advantage of Coveo ML ART, members with the required privileges must first create ART models.

Prerequisites

Before creating an ART model, your Coveo organization must collect enough data. To start providing recommendations, the model needs:

To analyze whether the search interface in which you want to integrate the ART model produces enough data, see Troubleshoot ART models.

Create an ART model

  1. On the Models (platform-ca | platform-eu | platform-au) page of the Coveo Administration Console, click Add model, and then click the Automatic Relevance Tuning card.

  2. Click Next.

  3. (Optional) In the Learning interval section, you can change the default and recommended values for both Data period and Building frequency.

  4. Click Next.

  5. (Optional) If you want your ART model to automatically consider cart and purchase event data, activate the Include commerce events option.

    Tip
    Leading practice

    Activate this option in Coveo for Commerce implementations, such as when configuring an ART model for product listing pages (PLPs).

    By default, ART models only consider the most-clicked products for a given query, from a specific search interface and tab combination. With this option enabled, they also consider all "add to cart" and "purchase" events that occur within a particular site (or application).

    However, ART models can’t learn from commerce events alone. These events can only help reinforce popularity trends and boost product rankings. ART models always require search and click events.

  6. (Optional) In the Apply filters on dataset section, you can add filters to refine the data that the model uses to make its recommendations. By narrowing down the dataset that a model uses, you can better customize relevancy for specific user groups and use cases. You can apply filters on all events, or on every event that belongs to a specific category, such as search, click, view, or custom events.

    Example

    Your Community search and Agent search have very specific vocabulary. You don’t want them to influence one another in the ART model learning process, so you add a filter on the Origin 1 (Page/Hub) dimension.

    This allows your model to learn independently from the actions performed on the different search interfaces.

    Tip

    The Data volume preview section shows the impact of filters on the data that’s available to train the model. This section includes data gathered only during the last seven days and doesn’t consider the selections from the Learning interval section.

    1. In the Select a dimension dropdown menu, select the dimension on which you want to base the learning of the model.

    2. In the Select an operator dropdown menu, select the appropriate operator.

    3. In the Select value(s) dropdown menu, add, type, or select the appropriate value.

    4. You can optionally add other filters by clicking Add.

      Note

      When configuring an ART model that doesn’t serve a commerce interface, don’t configure filters based on Custom dimensions as ART models don’t use custom events in their learning datasets. Doing so would result in the model not applying any filtering.

  7. Click Next.

  8. In the Name your model input, enter a meaningful display name for the model.

  9. (Optional) Use the Project selector to associate your model with one or more projects.

  10. Click Start building.

    Note

    On the Models (platform-ca | platform-eu | platform-au) page, under the Status column, in the model row, the value is most probably Inactive.

    The model value will change to Active when the model creation is complete (typically within 30 minutes, depending on the amount of usage analytics data to process). The model can only return recommendations when its status is Active.

    For more information on Coveo ML model statuses, see the Status column reference.

  11. (Optional) Configure advanced settings for your model.

  12. Associate the model with a pipeline to use the model in a search interface and test your model to make sure it behaves as expected.

    Important

    Make sure to follow the model association leading practices before associating your model with your production query pipeline.

Advanced configuration

You can configure advanced settings for your model to suit specific use cases.

Some advanced settings are available from the model options in the Administration Console, while others are only available through JSON configuration parameters.

Notes

To specify advanced settings

  1. On the Models (platform-ca | platform-eu | platform-au) page, click the model for which you want to specify advanced settings.

  2. Do one of the following:

  1. Click Save.

Manage an ART model

You can edit using the model options or JSON, delete, or review information for your model.

Edit an ART model

  1. On the Models (platform-ca | platform-eu | platform-au) page, click the model you want to edit, and then click Edit in the Action bar.

  2. On the subpage that opens, select the Configuration tab.

  3. Under Name, you can optionally edit the model’s display name.

  4. (Optional) Use the Project selector to associate your model with one or more projects.

  5. (Optional) Depending on whether you want your ART model to automatically consider cart and purchase event data, you can activate or deactivate the Include commerce events option.

    Tip
    Leading practice

    Activate this option in Coveo for Commerce implementations, such as when configuring an ART model for product listing pages (PLPs).

    By default, ART models only consider the most-clicked products for a given query, from a specific search interface and tab combination. With this option enabled, they also consider all "add to cart" and "purchase" events that occur within a particular site (or application).

    However, ART models can’t learn from commerce events alone. These events can only help reinforce popularity trends and boost product rankings. ART models always require search and click events.

  6. (Optional) In the Learning interval section, you can change the default and recommended values for both Data period and Building frequency.

  7. (Optional) In the Apply filters on dataset section, you can add filters to refine the data that the model uses to make its recommendations. By narrowing down the dataset that a model uses, you can better customize relevancy for specific user groups and use cases. You can apply filters on all events, or on every event that belongs to a specific category, such as search, click, view, or custom events.

    Example

    Your Community search and Agent search have very specific vocabulary. You don’t want them to influence one another in the ART model learning process, so you add a filter on the Origin 1 (Page/Hub) dimension.

    This allows your model to learn independently from the actions performed on the different search interfaces.

    Tip

    The Data volume preview section shows the impact of filters on the data that’s available to train the model. This section includes data gathered only during the last seven days and doesn’t consider the selections from the Learning interval section.

    1. In the Select a dimension dropdown menu, select the dimension on which you want to base the learning of the model.

    2. In the Select an operator dropdown menu, select the appropriate operator.

    3. In the Select value(s) dropdown menu, add, type, or select the appropriate value.

    4. You can optionally add other filters by clicking Add.

      Note

      When configuring an ART model that doesn’t serve a commerce interface, don’t configure filters based on Custom dimensions as ART models don’t use custom events in their learning datasets. Doing so would result in the model not applying any filtering.

  8. (Optional) Configure advanced settings for your model.

  9. Click Save.

    Note

    Some configuration changes initiate an automatic model rebuild when you save the model. The Models (platform-ca | platform-eu | platform-au) page shows your model’s current Status. Model settings take effect only when its status is Active.

    For more information on Coveo ML model statuses, see the Status column reference.

  10. Associate the model with a pipeline to use the model in a search interface and test your model to make sure it behaves as expected.

    Important

    Make sure to follow the model association leading practices before associating your model with your production query pipeline.

Edit an ART model JSON configuration

  1. Access the Models (platform-ca | platform-eu | platform-au) page.

  2. Click the desired model, and then click More > Edit JSON in the Action bar.

  3. In the Edit a Model JSON Configuration panel that appears, modify the existing model configuration:

    {
  "modelDisplayName": "<MODEL_DISPLAY_NAME>",
  "exportPeriod": "<EXPORT_PERIOD>",
  "intervalTime": <INTERVAL_TIME>,
  "intervalUnit": "<INTERVAL_UNIT>",
  "commonFilter": "<COMMON_FILTER>",
  "customEventFilter": "<CUSTOM_EVENT_FILTER>",
  "exportOffset": "<EXPORT_OFF_SET>",
  "searchEventFilter": "<SEARCH_EVENT_FILTER>",
  "viewEventFilter": "<VIEW_EVENT_FILTER>",
  "extraConfig": [<ADVANCED_ML_PARAMETERS>],
}

    Where:

    • modelDisplayName (string) is the name of the model appearing on the Models (platform-ca | platform-eu | platform-au) page.

    • exportPeriod (ISO-8601 string, required) is the period defining the age of the Coveo Analytics data used to build the model. Must be in the ISO8601 period format (that is, PyYmMwWdDThHmMsS).

      Note

      Unless an exportOffset is specified, the exportPeriod uses the moment when the model was generated as a base.

    • intervalTime (integer, required) is the number of intervalUnit (that is, DAY, WEEK, or MONTH) between each update of the model. Must be between 1 and 30 inclusively.

    • intervalUnit (string enum, required) is the duration unit of the interval between each update of the model. See intervalTime. Accepted values are: DAY, WEEK, and MONTH.

    • commonFilter (string) is the filter to apply to the common event dimensions (shared by all event types) in the export. Multiple filter parameters are joined with the AND operator.

    • customEventFilter (string) is the filter to apply to the custom event dimensions in the export. Multiple filter parameters are joined with the AND operator.

    • exportOffset (ISO-8601 string) is the offset of the usage analytics data used to build the model. Must be in the ISO8601 period format (that is, PyYmMwWdDThHmMsS). The default value is PT0S, meaning that all events are considered when building a model (the exportPeriod is based on the moment the model was generated).

      Example

      You want to ignore events that occur on the current day, so you set the exportOffset value to P1D.

    • searchEventFilter (string) is the filter to apply to the click and search event dimensions in the export. Multiple filter parameters are joined with the AND operator.

    • viewEventFilter (string) is the filter to apply to the view event dimensions (shared by all event types) in the export. Multiple filter parameters are joined with the AND operator.

    • extraConfig (array of string) are additional advanced parameters used to tailor the model to your use case.

  4. Click Save to apply your changes.

Delete an ART model

Note

If the model is associated with a query pipeline, make sure to dissociate the model from the query pipeline after deleting it.

  1. On the Models (platform-ca | platform-eu | platform-au) page, click the ML model that you want to delete, and then click More > Delete in the Action bar.

  2. In the panel that appears, click Delete.

Review model information

On the Models (platform-ca | platform-eu | platform-au) page, click the desired model, and then click View in the Action bar. For more information, see Reviewing model information.

Reference

ART models reference

Default number of recommended results: 5

Note

ART recommends the 5 best learned search results, but may recommend fewer than 5 when:

  • The query isn’t very frequent, and less than 5 items were learned for that query.

  • Some recommendations are secured and not accessible to the user performing the query.

  • Some recommendations are filtered out because they don’t match filters such as the current facet selections.

Facts

  • Following an empty query, ART returns the most clicked items during the data period of the model.

  • ART can inject items that wouldn’t normally be included in search results because they were learned to be relevant even if they don’t contain some or all of the searched keywords. This is one of the key ART benefits as a user can find the most useful items without having to type the right keywords or the specific synonym contained these items. This is the default behavior (the Match the Query parameter default value is false).

  • ART currently ignores all special characters or operators in the user entered query to only keep the keywords and therefore ignores the special behavior described in Using Special Characters in Queries or Search Prefixes and Operators.

  • This ART behavior may lead to unexpected search results when users want to take advantage of more advanced query syntax.

Example

A user is searching for items containing a specific phrase by entering the phrase enclosed in double quotes (see Searching for a phrase) and no items contain the phrase. The user would expect to get no search results, but ART can inject items that were clicked for similar queries made of one or some of the searched keywords (not in a phrase search).

Advanced model options

You can use the following options when configuring advanced settings for your model.

Suggestion filters

Use the Suggestion filters advanced setting to set the Coveo Analytics dimensions to be used as filters for potential suggestions.

The model filters the suggestions so that only the interaction data from events that include the specified dimensions are considered.

By default, the Search hub and Tab dimensions are selected. This means that for a query that originates from a specific tab in a search hub, only interactions recorded in the same search hub tab will be used by the model.

Example

Given the default values of Search hub and tab, and the following implementation:

  • Two Search hub values: partnerHub and techSupportHub

  • Four Tab values: all, documentation, training, and community

A total of eight possible filters will be created ( partnerHub/all, techSupportHub/all, partnerHub/documentation, etc.). This means that if a query originates from partnerHub/all, only interactions recorded in partnerHub/all will be used by the model to make suggestions for that query.

You can set this option to use the recommended Search hub and Tab dimensions (default), to not use filters at all, or to use custom filters based on dimensions of your choice.

To configure the suggestion filters

  1. On the Models (platform-ca | platform-eu | platform-au) page, click the model that you want to modify, and then click Edit in the Action bar.

  2. On the subpage that opens, select the Advanced tab.

  3. In the left menu of the Advanced tab, select Suggestion filters.

  4. Choose one of the following options:

    • Search hub and Tab (default): The model uses the Search hub and Tab dimensions as filters for suggestions.

    • No dimension filters: The model doesn’t use any filters for suggestions. This provides the same relevance across all search hubs that use the model. It’s useful when your model serves different search hubs in which the same source items are available.

    • Custom filters: The model uses the dimensions that you select as filters for suggestions.

      1. Optionally, select one or both of the Search hub and Tab filters.

      2. Under Other dimensions, select the dimensions you want to use to filter the model suggestions (for example, country and language).

        Important

        If you set a dimension other than the two default ones (Search Hub and Tab), you must also add the dimension at query time using the filters mlParameter.
        Custom filters configuration

  5. Click Save.

Test configuration mode

Note

The Test configuration mode advanced option is available only for sandbox organizations.

Sandbox organizations typically lack the amount of usage analytics data that’s required to train a model. The Test configuration mode option lets you build a model in a sandbox organization with little or infrequent usage analytics data so you can test the model.

When activated, this option reduces the amount of analytics data that’s required to build the model. It also reduces other frequency thresholds that discard queries or clicks that weren’t performed frequently enough.

Note

The usage of certain frequency thresholds, or the selection of a specific value for these frequency thresholds depends on the configuration and implementation of the model. As the possible combinations of threshold configurations are adapted for each model, these frequency thresholds aren’t listed in this section.

To activate the test configuration mode

  1. On the Models (platform-ca | platform-eu | platform-au) page, click the model for which you want to activate the test configuration mode, and then click Edit in the Action bar.

  2. On the subpage that opens, select the Advanced tab.

  3. In the left menu of the Advanced tab, select Test configuration mode.

  4. Select the Activate test configuration mode checkbox.

  5. Click Save.

Advanced JSON model parameters

You can use the following JSON parameters when configuring advanced settings for your model.

automaticContextDiscovery (boolean)

Sets whether the model evaluates custom usage analytics dimensions prefixed with context_ to provide predictions or recommendations.

Default: true

When set to false, the model doesn’t automatically consider user context found in data. However, it will use user context fields defined in the userContextFields parameter.

To set the parameter

Access the model’s JSON editor, and then add the parameter configuration to the extraConfig object.

Example

You want to build a model that doesn’t evaluate custom usage analytics dimensions prefixed with context_. Therefore, you enter the following configuration in the extraConfig object:

"extraConfig": {
    "automaticContextDiscovery": false
}

commerceSupport (object)

Whether the ART model should consider cart and purchase events to boost products on a commerce results page.

Default: false

When set to true, the ART model considers products that have been either purchased or added to a cart following a given query as items to be boosted for that particular query and automatically assign the required weight to each event.

To set the parameter

Access the model’s JSON editor, and then add the parameter configuration to the extraConfig object.

Example

You want your ART model to consider cart and purchase events to boost products on a commerce results page.

Therefore, you enter the following configuration in the extraConfig object:

"extraConfig": {
  "commerceSupport": {
    "enabled": true
  },
}

filterFields (list of strings)

Use this parameter to set the Coveo Analytics dimensions to be used as filters for potential suggestions.

The model filters the suggestions so that only the interaction data from events that include the specified dimensions are considered.

Default: ["originLevel1", "originLevel2"].

Note

Given the default value of ["originLevel1", "originLevel2"], and the following implementation:

  • Two originLevel1 values: partnerHub and techSupportHub

  • Four originLevel2 values: all, documentation, training, and community

A total of eight possible filters will be created ( partnerHub/all, techSupportHub/all, partnerHub/documentation, etc.). This means that if a query originates from partnerHub/all, only interactions recorded in partnerHub/all will be used by the model to make suggestions for that query.
Important

If you set a field other than the two default ones (originLevel1 and originLevel2), you must also add the field at query time using the filters mlParameter.

To set the parameter

Access the model’s JSON editor, and then add the parameter configuration to the extraConfig object.

Example

You want your ML model to consider the possible value combination of the originContext and originLevel2 dimensions when filtering results because some of the results are not available in some other combinations.

Therefore, you enter the following configuration in the extraConfig object:

"extraConfig": {
    "filterFields": [
      "originContext",
      "originLevel2"
    ]
}

This would require sending the dimension values at query time in the filters mlParameter as follows:

"mlParameters": {
    "filters": {
          "originContext": "<MY-CONTEXT-VALUE>",
          "originLevel2": "<TAB-VALUE>"
    }
}

Moreover, you may want to build a model that doesn’t use filters at all since all items are accessible everywhere. You can do so by setting the filterFields parameter to empty in a model configuration. This allows you to provide the same relevance across all search hubs using the model.

For example:

"extraConfig": {
    "filterFields": []
}

recommendProductGroup (boolean)

Sets whether the model recommends product groups rather than individual items.

When set to true, the model uses grouping fields to identify the content to recommend, ignoring the items' unique content ID (for example, permanentid).

To set the parameter

Access the model’s JSON editor, and then add the parameter configuration to the commerceSupport object in extraConfig.

Example

You want your model to use groups of products rather than individual items when providing recommendations.

Therefore, you enter the following configuration in the commerceSupport object:

"extraConfig": {
    "commerceSupport": {
      "enabled": true,
      "recommendProductGroup": true
    }
  }

userContextFields (list of strings)

The usage analytics dimensions whose values should be used as the user context by the model to influence the ranking scores of items.

Important

When configuring the userContextFields advanced parameter, make sure that the related dimension values are sent at query time in the context parameter.

To set the parameter

Access the model’s JSON editor, and then add the parameter configuration to the extraConfig object.

Example

You want to build an ML model that uses the originLevel3 and userGroups usage analytics dimensions as the user context to influence the ranking scores of items.

Therefore, you enter the following configuration in the extraConfig object:

"extraConfig": {
    "userContextFields": [
      "originLevel3",
      "userGroups"
    ]
}

"Status" column

On the Models (platform-ca | platform-eu | platform-au) page of the Administration Console, the Status column indicates the current state of your Coveo ML models.

The following table lists the possible model statuses and their definitions:

Status Definition Status icon

Active

The model is active and available.

Active

Build in progress

The model is currently building.

Building

Inactive

The model isn’t ready to be queried, such as when a model was recently created or the organization is offline.
Click See more details for additional information (see Review model information).

Inactive

Limited

Build issues exist that may affect model performance.
Click See more details for additional information (see Review model information).

Limited

Soon to be archived

The model will soon be archived because it hasn’t been queried for an extended period of time.
Click Delete to remove the model.
Learn more about archived models.

Archive pending

Error

An error prevented the model from being built successfully.
If it’s a temporary system error, check back soon. Otherwise, click See more details for additional information (see Review model information).

Error

Archived

The model was archived because it hasn’t been queried for an extended period of time.
Click Delete to remove the model.
Learn more about archived models.

Archived

"Learning interval" section

In this section, you can modify the following:

  • Data period: The usage analytics data time interval on which the model will be based.

  • Building frequency: The rate at which the model is retrained.

Set the Coveo ML model training Building frequency based on the Data Period value. Less frequent for a larger Data Period and more frequent for a smaller Data Period as recommended in the following table.

Data period

Building frequency

Daily

Weekly

Monthly

1 month

check

check

3 months (Recommended)

check

6 months

check

The more data the model has access to and learns from, the better the recommendations. As a general guide, a usage analytics dataset of 10,000 queries or more typically allows a Coveo ML model to provide very relevant recommendations. You can look at your Coveo Analytics data to evaluate the volume of queries on your search hub, and ensure that your Coveo ML models are configured with a training Data period that corresponds to at least 10,000 queries. When your search hub serves a very high volume of queries, you can consider reducing the data period so that the model learns only more recent user behavior and be more responsive to trends.

A Coveo ML model regularly retrains on a more recent Coveo UA dataset, as determined by the Building frequency and Data period settings, to ensure that the model remains up-to-date with the most recent user behavior.

Note

If you’re testing the model in a sandbox environment in which very little analytics data is available to train the model, you can activate the Test configuration mode advanced option to ensure the model provides recommendations.

Required privileges

The following table indicates the privileges required to use elements of the Models page and associated panels (see Manage privileges and Privilege reference).

Action Service - Domain Required access level

View models

Machine Learning - Models
Organization - Organization
Search - Query pipelines

View

Edit models

Organization - Organization
Search - Query pipelines

View

Machine Learning - Models

Edit