AI Assistant Recommendations

The AI Assistant Recommendations feature suggests courses that may interest learners by analyzing their training history. The recommendations are indicated by a star rating based on the level of relevance to the learner. When activated, learners can see recommended courses in the AI Assistant Recommendations page in the Learning Center.

Courses from the learner's training history are used to retrieve associated courses from two areas:

  • Courses explicitly configured by a training administrator using the Catalog Editor. There may be zero or more courses when assembling the recommendation list, but all those present are assumed to be of high quality and thus receive a rating of 1.0.
  • Courses ranked and associated with courses in the learner's training history by the background AI task running once per week. Each course may have an association rating between 0.0 and 1.0.

The system only displays courses that are in catalogs that the learner has permission to view. 

A Recommendations widget is available that allow learners to view AI Assistant Recommendations.

If the Show only "Active" sessions System Configuration setting is enabled, it will apply to the AI Recommendations and only active courses will display.

It is possible that the learner's training history or selected course does not have sufficient information to warrant a recommendation. In this case, the learner is presented with the following message:

Sorry, but there is not yet enough interaction and training analysis to provide you with meaningful course recommendations.

Uses of AI Assistant Recommendations

Users have several options available to refine the recommendations, which are themselves presented with a 'star' rating indicating the most relevant recommendations and an active link on each course title to take the user to the corresponding catalog page.  

Recommendation Focus

  • Self-Enrolled Courses - the recommendations will be based on courses in the user's training history that were self-enrolled.
  • Assigned Courses - the recommendations will be based on courses in the user's history that were assigned (e.g. group enroll, auto-enroll,...).
  • Random Recommendations - the recommendations will be based on all courses in the user's history, and since each course can have multiple associated courses, the assistant will use a probability function to vary the returned list instead of simply ranking the recommended courses and returning the highest-ranked courses (which is what happens with the first two options above).  This means that this list can vary each time a "surprise me" request is made.

Additional items, if configured:

  • Job Title - the recommendations will be based on courses in the user's history compared to the history of other users with the same job title
  • Organization Unit - the recommendations will be based on courses in the user's history compared to the history of other users with the same organization unit (based on the highest level visible for the user’s role).
  • User Groups - the recommendations will be based on courses in the user's history compared to the history of other users in one or more of the same user groups as the logged-in user.
  • Job Profiles - the recommendations will be based on courses in the user's history compared to the history of other users assigned one or more of the same job profile(s).
  • Competencies - the recommendations will be based on courses in the user's history compared to the history of other users awarded one or more the same competencies.
  • Certifications - the recommendations will be based on courses in the user's history compared to the history of other users with one or more of the same certifications. 

Courses to Consider

  • Consider ALL My Training - all courses in the user's history will be used in the analysis. This is the  default case.
  • Consider Only Recently Completed Training (30 Days) - only courses completed in the past month will be used for the analysis.
  • <Individual Courses listed> - the remainder of the drop-down options will list individual courses in the user's  history so that they can select an individual course for recommendations.

Display Limit

The number of recommendations returned is limited to the number selected in this field.  Typically, these are returned in the order of relevance.  The exception is the Random Recommendations focus.

Additional Recommended Course Information

Course Language

Courses with language bundles in the language selected by the logged-in user will display.  Courses without language bundles in the selected language display the primary language bundle.  This is the same functionality as Catalog Browse and Catalog Search.

Catalog Visibility

Only courses visible to the learner via Search or Browse appear for Recommendations.  These are courses that are configured to show in a catalog and assigned to catalogs that the user has permission to view.

AI Assistant Recommendations Ratings Task

The AI Assistant Recommendations Rating background task uses a Collaborative Filter algorithm to rate relationships between courses. It runs once per week and the ratings are updated also only once per week. You can, however, run the task on an ad hoc basis from the Scheduled Tasks page in the Manage Center.

The task analyses all of the training history in the LMS. A typical analysis of a large database containing five million rows of training history records can be analysed in five minutes. This can temporarily use 500 MB to 1,000 MB of RAM, which is freed once the results are stored back in the database.

By default, only the past 18 months of training data are analyzed and each course must have at least 10 enrollments to be included in the recommendations.  Both of these parameters are configurable.  

Configure AI Assistant Recommendations

There are system configurations available for the AI Assistant Recommendations.

Transcript Data - Timeline for Training Data

By default, only the past 18 months of training data are analyzed.  A System Configuration setting allows you to change this parameter:

  1. Go to Manage Center > System > System > System Configuration.
  2. In the Select a Category field, select Records/Transcript.
  3. Locate the Transcript history period (months) field. Change the value to the desired timeline in months (minimum 6).

Transcript Data - Minimum Course Enrollment

By default, the system will require each course to have at least 10 enrollments to be included in the recommendations.  A System Configuration setting allows you to change this parameter:

  1. Go to Manage Center > System > System > System Configuration.
  2. In the Select a Category field, select Records/Transcript.
  3. Locate the Minimum enrollments per module field. Change the value to the desired number of enrollments (minimum 5).

Recommendation Options

By default, the AI Assistant Recommendations functionality only looks at courses on the transcript, and is not associated with user metadata.  This system configuration to determine which user data elements allows transcript comparison with other users.

  • Job Title - Courses based on skills associated with the user’s job title.
  • Organization Unit - Courses taken by other users within the same organization unit (based on the highest level visible for the user’s role).
  • User Groups - Courses taken by other users in the same user groups as the logged-in user.
  • Job Profiles - Courses based on the histories of users with the same job profile(s).
  • Competencies - Courses based on the histories of users with the same awarded, unexpired competencies.
  • Certifications - Courses based on the histories of users with the same certifications.

To configure the setting:

  1. Go to Manage Center > System > System > System Configuration.
  2. In the Select a Category field, select Records/Transcript.
  3. Locate the field and configure it as necessary.