Item-Item Similarities

Item-item similarities are similarities between items. Fusion computes similarities based on user-item preferences. You can use item-item similarities for whatever you want. The most common use is as input data for Spark-based collaborative filtering.

By default, the <collection>_item_recommendations job is configured to produce item-item similarities and item-item recommendations.

To produce item-item similarities and item-item recommendations:

  1. From the Fusion Launcher, click Search > Home Home > Jobs.

  2. Click the <collection>_items_for_user_recommendations job.

  3. In the Item-to-item Similarity Collection text box, enter <collection>_items_for_item_recommendations.

  4. Click Save to save the job configuration.