With the serendipity factor you can tune your algorithm to re-rank the output and thereby optimize the recommendations on your website by choosing the optimal balance between favoring the most popular products and a greater diversity of products in the product recommendations.
The serendipity tuning is most useful for modules as “GetRelatedItems”, “GetSimilarItems” and “GetRelatedItemsForBasket”.
The Serendipity score is a factor to adjust the balance of recommended products between popular products or diversity of products. When adjusting the serendipity score, you can control to what extend the most popular products are to be shown in the recommendations on various product pages.
A low serendipity value (0-1500) favors the most popular products, and a high value (5000) favors more diversity in the recommended products across all products on the website. The default value is 3000. However, it is crucial to test the output with various serendipity scores to find the most optimal setup for each business.
You choose the most optimal serendipity score in each module in the Raptor Control Panel. Go to Recommendation in the top right corner à choose products à click on “Raptor Web Advisor” à choose the module you want to adjust. Here you can test the output when adjusting the serendipity score, which you adjust in the box “Optional Input Parameters” - “Module Tuning”.
Please sign in to leave a comment.