Cluster-based assessment of user interests

This invention describes computer methods for organizing a user's past interactions, like purchases or views, into groups (clusters) based on how similar the items are. This grouping helps a computer system better understand the user's interests, which can then be used to select better items to recommend. For instance, the system can display these clusters to the user, allowing them to rate or tag entire groups of items at once, further refining future recommendations. The claims specifically focus on using a computer to apply a clustering algorithm to a user's collection of items, such as their purchase history, to subdivide them into these clusters.