A method of adaptively updating a first recommended user group list of a first user connected to a network, in which a predetermined number of second users are selected from the first recommend user group list that is a list of users having a high similarity to the first user in consuming contents. A predetermined number of third users having a high similarity to the first user is selected from second recommend user group lists respectively of the selected second users possess. The first recommend user group list is updated to include the selected third users. As the preference of a user changes, the recommend user group may be reconfigured with updated recommend users by reflecting a corresponding preference. Also, quality contents can be provided by recommending a user of a corresponding terminal with the contents preferred by other users in the updated recommend user group list.