US Patent 11556966 Item-to-item recommendations

An approach is disclosed for providing item-to-item recommendations. The approach receives data for an item. The approach generates recommended candidate item data for the data of the received item. The approach generates feature data of the recommended candidate item data and the data of the received item. The feature data may include one or more similarities between the data of the received item and data corresponding to a respective recommended candidate item. The approach generates ranked recommended candidate item data based on the generated feature data and at least one weight determined for candidate test item data. The candidate test item data may include data corresponding to the recommended candidate item data.