Recommendation strategies are the default and preferred way of generating recommendation data. If possible, use a recommendation strategy instead of a recommendation configuration when creating a new recommendation.
The first step in creating a recommendation is to define what type of recommendation you want to create and what specifically you want to include in the recommendation, that is, what it is that you want to recommend. You do this with a recommendation configuration.
Figure: Recommendation configuration in the Frosmo Control Panel
The recommendation configuration defines the logic and settings for generating the recommendation data. The configuration consists of:
- Name and description
- Recommendation type
- Frequency of regenerating the recommendation data
- Number of past hours of source data on which to base the recommendation
- Maximum number of recommended items to return
- Setting for mapping the generated recommendation data to source data, such as product data
- Settings for filtering the generated recommendation data
Based on the recommendation configuration, the Frosmo Recommendations system retrieves the necessary usage data from the Frosmo Data Pipeline, and applies the requested algorithm to generate the appropriate recommendation data. You fetch the generated data with the Recommendations API, and you then use the data in a modification to build and display the final recommendation. If you use a template for the modification content, you fetch and use the recommendation data in the template, and then use the template in the modification.
Figure: Recommendation configuration in the recommendation workflow (click to enlarge)
Where to go next
- To start developing recommendations, see Creating recommendations with recommendation configurations.
- To learn how to create and manage configurations in the Frosmo Control Panel, see Working with recommendation configurations (Frosmo Control Panel).
- If you're creating your own Frosmo application, and you want to learn how to create and manage recommendation configurations with the Graniitti API, see Working with recommendation configurations (Graniitti API).
- For the full description of the recommendation configuration data model, see Recommendation configuration reference.
- For a practical example of creating from scratch a recommendation that uses a recommendation configuration, see Example: Recommending products purchased together (configuration).