Machine Learning Advanced Parameters

When creating or updating a Coveo Machine Learning (Coveo ML) model, you can specify various advanced parameters to tailor the model to specific use cases. This article provides reference information on the available advanced model parameters.

Notes

In addition to the custom model parameters described in this article, you can also use the mlParameters query parameter to adjust the way your Coveo ML models are used at query time.

Specify Advanced Parameters

  1. On the Models (platform-eu | platform-au) page, click the model for which you want to add advanced parameters, and then, in the Action bar, click Edit.

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

  3. At the upper-left corner, select the advanced configuration parameter that you want to configure for this model. Available parameters differ depending on the type of model you want to configure. Select below the type of model for which you want to configure advanced parameters for instructions about the available advanced parameters:

Automatic Relevance Tuning (ART) Advanced Model Parameters

Overrides

By default, a Coveo ML model evaluates whether to use certain context dimensions in two phases:

  1. Based on a statistical analysis, the user context selection algorithm filters out irrelevant dimensions.

  2. During its training phase, the model then evaluates whether the remaining dimensions should be used to improve the model’s output.

The Overrides advanced model parameter lets you override the user context selection algorithm by specifying dimension key names that must either bypass the algorithm filtering phase, or be completely ignored by both the algorithm and the model.

Examples
  • You want the c_context_brand and c_context_contact_primary_role dimension keys to bypass the user context selection algorithm’s filtering phase, forcing them to be considered by the model during its training phase. Therefore, you specify them in the Dimensions to bypass the algorithm section when configuring the Overrides advanced model parameter.

  • You want the c_context_brand and c_context_contact_primary_role dimension keys to be completely ignored by both the user context selection algorithm and the model. Therefore, you specify them in the Dimensions to ignore section when configuring the Overrides advanced model parameter.

To override the user context selection algorithm

  1. On the models page, click the model for which you want to specify user context dimensions that must bypass the user context selection algorithm, and then, in the Action bar, click Edit.

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

  3. In the upper-left corner, select Overrides.

  4. Depending on whether you want to specify user context dimensions that must bypass the user context selection algorithm filtering phase, or user context dimensions that must completely be ignored by both the algorithm and the model:

    • To specify user context dimensions that must bypass the algorithm filtering phase, in the Dimensions to bypass the algorithm section, select the desired user context dimensions.

    • To specify user context dimensions that must completely be ignored by both the algorithm and the model, in the Dimensions to ignore section, select the desired user context dimensions.

  5. Click Save.

Note

If the same context key is used in both the Dimensions to bypass the algorithm and Dimensions to ignore sections, the one specified in the Dimensions to bypass the algorithm section takes precedence.

Suggestion Filters

The Suggestion filters parameter allows you to select the Coveo Usage Analytics (Coveo UA) dimensions to be used as filters for potential suggestions.

An item will be suggested by the model only if it has been clicked with the specified filter values.

By default, the Search Hub and Tab values are selected.

Example

With the default values (i.e., Search Hub and Tab ), if there are two possible Search Hub values (e.g., partnerHub and techSupportHub) and four possible Tab values (e.g., 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 partnerHub/all is received at query time, only the items clicked in partnerHub/all will be returned by the model.

You may want to change the default model behavior by indicating the dimensions on which you want the model to filter the recommendations.

Important

If you set other fields than the two default ones (i.e., Search Hub and Tab), you must also add the values at query time using the filters mlParameters.

To configure custom filters

  1. On the models page, click the model for which you want to configure custom filters, and then, in the Action bar, click Edit.

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

  3. In the upper-left corner, select Suggestion filters.

  4. Optionally, deselect the default Search Hub and Tab filters.

  5. Select the Custom filter dimensions checkbox.

  6. In the input box that opens, select the dimensions combination you want to use to filter the model recommendations (e.g., country and language).

  7. Click Save.

Configure a Model Without Filters

You may want to build a model that doesn’t use filters at all to provide the same relevance across all search interfaces using the model.

This can be useful when your model serves search interfaces in which the same source items are available.

You can do so by deselecting the default Search Hub and Tab values.

no filters configuration

Test Configuration Mode

When activated, the Test configuration mode parameter reduces the amount of analytics data required to build the model. It also reduces other frequency thresholds that discard queries or clicks that were not performed frequently enough.

Note that 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.

Important

This parameter should be used in sandbox environments, when very few analytics are available to train a model.

To activate the test configuration mode

  1. On the models page, click the model for which you want to activate the test configuration mode, and then, in the Action bar, click Edit.

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

  3. At the upper-left corner, select Test configuration mode.

  4. Select the Activate test configuration mode check box.

  5. Click Save.

Query Suggestions (QS) Advanced Model Parameters

Overrides

By default, a Coveo ML model evaluates whether to use certain context dimensions in two phases:

  1. Based on a statistical analysis, the user context selection algorithm filters out irrelevant dimensions.

  2. During its training phase, the model then evaluates whether the remaining dimensions should be used to improve the model’s output.

The Overrides advanced model parameter lets you override the user context selection algorithm by specifying dimension key names that must either bypass the algorithm filtering phase, or be completely ignored by both the algorithm and the model.

Examples
  • You want the c_context_brand and c_context_contact_primary_role dimension keys to bypass the user context selection algorithm’s filtering phase, forcing them to be considered by the model during its training phase. Therefore, you specify them in the Dimensions to bypass the algorithm section when configuring the Overrides advanced model parameter.

  • You want the c_context_brand and c_context_contact_primary_role dimension keys to be completely ignored by both the user context selection algorithm and the model. Therefore, you specify them in the Dimensions to ignore section when configuring the Overrides advanced model parameter.

To override the user context selection algorithm

  1. On the models page, click the model for which you want to specify user context dimensions that must bypass the user context selection algorithm, and then, in the Action bar, click Edit.

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

  3. In the upper-left corner, select Overrides.

  4. Depending on whether you want to specify user context dimensions that must bypass the user context selection algorithm filtering phase, or user context dimensions that must completely be ignored by both the algorithm and the model:

    • To specify user context dimensions that must bypass the algorithm filtering phase, in the Dimensions to bypass the algorithm section, select the desired user context dimensions.

    • To specify user context dimensions that must completely be ignored by both the algorithm and the model, in the Dimensions to ignore section, select the desired user context dimensions.

  5. Click Save.

Note

If the same context key is used in both the Dimensions to bypass the algorithm and Dimensions to ignore sections, the one specified in the Dimensions to bypass the algorithm section takes precedence.

Query Suggestions Format

The Query suggestions format advanced parameter allows you to find specific patterns and reformat them in queries suggested by a Coveo ML QS model.

Example

You want your QS model to reformat 5551234567 to 555-123-4567.

Therefore, you configure the Query suggestions format advanced parameter as follows:

query suggestion pattern

To configure the query suggestions format parameter

  1. On the models page, click the QS model for which you want to configure the query suggestions format parameter, and then, in the Action bar, click Edit.

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

  3. At the upper-left corner, select Query suggestions format.

  4. In the Enter the regular expression needed to find a pattern in query suggestions input, enter a regular expression that matches the pattern to find in query suggestions.

  5. In the Replace by input, enter a replacement pattern to apply when a query suggestion matches the regex entered in the Regular expression matching the pattern to find input.

  6. Click Save.

Suggestion Filters

The Suggestion filters parameter allows you to select the Coveo Usage Analytics (Coveo UA) dimensions to be used as filters for potential suggestions.

An item will be suggested by the model only if it has been clicked with the specified filter values.

By default, the Search Hub and Tab values are selected.

Example

With the default values (i.e., Search Hub and Tab ), if there are two possible Search Hub values (e.g., partnerHub and techSupportHub) and four possible Tab values (e.g., 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 partnerHub/all is received at query time, only the items clicked in partnerHub/all will be returned by the model.

You may want to change the default model behavior by indicating the dimensions on which you want the model to filter the recommendations.

Important

If you set other fields than the two default ones (i.e., Search Hub and Tab), you must also add the values at query time using the filters mlParameters.

To configure custom filters

  1. On the models page, click the model for which you want to configure custom filters, and then, in the Action bar, click Edit.

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

  3. In the upper-left corner, select Suggestion filters.

  4. Optionally, deselect the default Search Hub and Tab filters.

  5. Select the Custom filter dimensions checkbox.

  6. In the input box that opens, select the dimensions combination you want to use to filter the model recommendations (e.g., country and language).

  7. Click Save.

Configure a Model Without Filters

You may want to build a model that doesn’t use filters at all to provide the same relevance across all search interfaces using the model.

This can be useful when your model serves search interfaces in which the same source items are available.

You can do so by deselecting the default Search Hub and Tab values.

no filters configuration

Test Configuration Mode

When activated, the Test configuration mode parameter reduces the amount of analytics data required to build the model. It also reduces other frequency thresholds that discard queries or clicks that were not performed frequently enough.

Note that 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.

Important

This parameter should be used in sandbox environments, when very few analytics are available to train a model.

To activate the test configuration mode

  1. On the models page, click the model for which you want to activate the test configuration mode, and then, in the Action bar, click Edit.

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

  3. At the upper-left corner, select Test configuration mode.

  4. Select the Activate test configuration mode check box.

  5. Click Save.

Content Recommendations (CR) Advanced Model Parameters

Test Configuration Mode

When activated, the Test configuration mode parameter reduces the amount of analytics data required to build the model. It also reduces other frequency thresholds that discard queries or clicks that were not performed frequently enough.

Note that 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.

Important

This parameter should be used in sandbox environments, when very few analytics are available to train a model.

To activate the test configuration mode

  1. On the models page, click the model for which you want to activate the test configuration mode, and then, in the Action bar, click Edit.

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

  3. At the upper-left corner, select Test configuration mode.

  4. Select the Activate test configuration mode check box.

  5. Click Save.

URL Format

The URL format advanced model parameter allows you to specify sets of patterns to find and reformat in URLs of recommended items.

Example

You want your CR model to remove trailing labels in URLs.

Therefore, you configure the URL format advanced parameter as follows:

url format

To configure the URL format advanced parameter

  1. On the models page, click the CR model for which you want to configure the URL format advanced parameter, and then, in the Action bar, click Edit.

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

  3. At the upper-left corner, select URL format.

  4. In the Regular expression matching the pattern to find input, enter a regular expression that matches the pattern to find in URLs of recommended items.

  5. In the Replace by input, enter a replacement pattern to apply when a URL matches the regex entered in the Regular expression matching the pattern to find input.

  6. Click Save.

Dynamic Navigation Experience (DNE) Advanced Model Parameters

Suggestion Filters

The Suggestion filters parameter allows you to select the Coveo Usage Analytics (Coveo UA) dimensions to be used as filters for potential suggestions.

An item will be suggested by the model only if it has been clicked with the specified filter values.

By default, the Search Hub and Tab values are selected.

Example

With the default values (i.e., Search Hub and Tab ), if there are two possible Search Hub values (e.g., partnerHub and techSupportHub) and four possible Tab values (e.g., 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 partnerHub/all is received at query time, only the items clicked in partnerHub/all will be returned by the model.

You may want to change the default model behavior by indicating the dimensions on which you want the model to filter the recommendations.

Important

If you set other fields than the two default ones (i.e., Search Hub and Tab), you must also add the values at query time using the filters mlParameters.

To configure custom filters

  1. On the models page, click the model for which you want to configure custom filters, and then, in the Action bar, click Edit.

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

  3. In the upper-left corner, select Suggestion filters.

  4. Optionally, deselect the default Search Hub and Tab filters.

  5. Select the Custom filter dimensions checkbox.

  6. In the input box that opens, select the dimensions combination you want to use to filter the model recommendations (e.g., country and language).

  7. Click Save.

Configure a Model Without Filters

You may want to build a model that doesn’t use filters at all to provide the same relevance across all search interfaces using the model.

This can be useful when your model serves search interfaces in which the same source items are available.

You can do so by deselecting the default Search Hub and Tab values.

no filters configuration

Test Configuration Mode

When activated, the Test configuration mode parameter reduces the amount of analytics data required to build the model. It also reduces other frequency thresholds that discard queries or clicks that were not performed frequently enough.

Note that 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.

Important

This parameter should be used in sandbox environments, when very few analytics are available to train a model.

To activate the test configuration mode

  1. On the models page, click the model for which you want to activate the test configuration mode, and then, in the Action bar, click Edit.

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

  3. At the upper-left corner, select Test configuration mode.

  4. Select the Activate test configuration mode check box.

  5. Click Save.

Product Recommendations (PR) Advanced Model Parameters

Test Configuration Mode

When activated, the Test configuration mode parameter reduces the amount of analytics data required to build the model. It also reduces other frequency thresholds that discard queries or clicks that were not performed frequently enough.

Note that 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.

Important

This parameter should be used in sandbox environments, when very few analytics are available to train a model.

To activate the test configuration mode

  1. On the models page, click the model for which you want to activate the test configuration mode, and then, in the Action bar, click Edit.

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

  3. At the upper-left corner, select Test configuration mode.

  4. Select the Activate test configuration mode check box.

  5. Click Save.

What's next for me?