This paper presents an open recommender system to ease the entering barriers due to lack of sufficient background knowledge for small or new service providers. The proposed Open Preference and Feature recommender (OPF) uses user preference and item feature as the basis of recommendations, since the generality of preference and feature and therefore meets the needs of an open recommender system.
In OPF, the group preference of similar tasted users and positive correlative features of items are taken into considerations to improve accuracy of recommendations. Since OPF uses a general preference of user for a class of items, it significantly reduces the space complexity to O(M+N). Simulations reveal that for even basing on general class preference, OPF obtains a low mean absolute error as 0.98 with coverage of 98.335.