Trusted Website Recommendation via Group Segmentation

The patent segments the global rating system into group-specific rating systems. Instead of using a single global popularity ranking that aggregates all ratings, the system divides ratings into separate groups based on social relationships (e.g., friends, family, colleagues). Each user receives recommendations based on ratings from their specific group members, thereby maintaining quantity of ratings while improving relevance to individual user tastes through targeted segmentation.

The patent applies local quality by differentiating the trustworthiness of ratings based on the relationship between the rater and the user. Instead of treating all ratings equally, the system assigns different levels of trust or weight to ratings from different groups. Ratings from trusted group members carry more weight and influence recommendations more strongly, while ratings from strangers have minimal impact. This allows the system to maintain broad coverage of rated sites while ensuring that recommendations are driven by trustworthy sources.

The patent introduces group members as intermediaries between external editors and users. Instead of users directly consuming editorial evaluations, the system uses trusted friends and contacts within social groups to evaluate and recommend content. These intermediaries act as a bridge, providing quality control through their own judgment while simultaneously personalizing recommendations based on their understanding of the user's preferences. This intermediary layer combines the benefits of quality control with personalization.