Temu is a personalized recommendation engine based on user behavior. It uses the user's historical behavior (such as purchases, ratings, browsing, etc.) and preferences to recommend products. To remove duplicate items, you can take the following measures:

Remove duplicate items: By integrating the user's shopping history and rating data, you can find duplicate items. Then remove these items from the recommendation list to avoid the same or similar items being recommended to the same user.

Deduplicate based on similarity: Temu can use collaborative filtering algorithms to find items that are similar to existing items and remove these items from the recommendation list.

Randomize recommendations: You can introduce random numbers into the system to decide which item will be recommended to the user to avoid recommending the same item frequently.

Recommend in context: Temu can use contextual information to better understand the user's interests and needs, and provide personalized de-duplicate recommendations accordingly. For example, if a user has browsed many similar products when purchasing a specific type of item, Temu can infer that the user may not be interested in this type of item