Design And Realization Of Personalized Coloring Ring Back Tone Recommendation System

As an emerging media, Mobile Phone has a good future in China. Value-added services of mobile communications business have been paid high attention by operators. CRBT (Coloring Ring Back Tones) is such a value-added business. Facing such a wide variety CRBTs, users can not listen to them one by one, so it is hard for them to find their favorite CRBTs, which affects users for the use of CRBT business to a certain extent. At present, operators only recommend the newest CRBTs or the top several CRBTs in ranking list, even give users free CRBT service to allow users to try simply. These ways above are difficult to arouse the interest of users really, but also are not able to demonstrate to users of personalized service. So offered personalized CRBT recommendation services, users can increase the interest and loyalty for CRBT. This is an inevitable requirement.This paper introduces the concept, development and the need for personalized service of CRBT, then gives the research condition of personalized recommendation and key techniques, finally does detailed analysis to content recommendation and collaborative recommendation algorithms. This paper presents the configuration of CRBT genome in which external attributes such as CRBT name, singer, published time, etc. are considered. In addition, this paper presents the analysis to internal attributes of CRBT, including lyric emotion type, rhyming degree and cantus emotion type of CRBT and so on. These genomes can describe many aspects of the characteristics of CRBT comprehensively and in detail. Then this paper explains the computing process of various gene values in detail and implements CRBT genome automatic tagging.On the basis of building CRBT genome, this paper creates the tone profiles of CRBT, then computes the interest values to the various gene values of CRBT according to the historical using recorded of users and related tone profiles. After that, user profiles are created by automatic modeling. On the basis of tone profiles and user profiles, this paper designs a personalized CRBT recommendation system, which uses the mixed recommendation method combined by content recommendation and collaborative recommendation and implements personalized CRBT recommendation to the target users.