How does it work?
The CEM engine extracts the most current active users (known as leaders or alpha-users) and clusters them into different communities based on their fixed profiles (descriptive information of the user) and their behaviors (purchase history, navigation behaviors, ratings, comments, etc.). All other users are distributed (and associated) into these communities based on their level of affiliation with the leaders. The engine determines the most current trends of each community by pre-calculating the most active content (products, news, ads, users, etc.) relevant for each community.
In real time, the engine will automatically classify new users to an existing community and deliver real-time recommendations of any kind. Recommendations are a threefold process based on the analysis of a segment of community trends combined with an evaluation of relevance based on the user’s profile (who are they) and the current context (what are they looking at now). Whether the user is a frequent user with a fully defined profile or a new anonymous user, the engine will seek the most appropriate recommendation strategy to optimize the quality of the recommendations.