Design And Implementation Of Color Ring Recommendation Module Of CRBT Professional Analysis System

Color-Ring-Back-Tone (CRBT) service has become one of the most important value-add services of telecommunication operators. But the user number of CRBT is meeting the limit, and the activity of users is decreasing. In order to attract the users to order color rings more frequently, telecommunication operators often starts campaigns to recommend some color rings to the users. But the traditional campaigns are of less precision, so the results are often not successful.It is the most popular recommendation method to recommend products based on association rules. But most of these recommendations are based on the single-dimensional association rules. This method has some limitations, for example, it could not emphasize on the newly-online rings and some special rings.The color ring recommendation module of CPAS (CRBT Professional Analysis System) solves these problems effectively. The module consists of ETL (Extract, Transform and Load) sub-module, association rule sub-module, recommendation weight adjust sub-module and recommendation sub-module. The module adopts multidimensional association rule algorithm to support the recommendation of newly-online rings. In the process of recommendation, the Recommendation Weight of the ring, which is decided by algorithms and can be modified by system user, is an important parameter. In this way, the special rings which need to be promoted could be emphasized during the recommendation.There are many types of recommendation in the module, for example, to recommend ring based on ring, to recommend ring based on user, to recommend user based on ring and etc. The module is able to support offline recommendation and online recommendation. In the process of offline recommendation, the module offers the precision marketing recommendation data to the other platform, e.g. SMS (Short Message Service) platform, to execute the campaign. In the process of online recommendation, CRBT platform could get recommendations via Web Services offered by the module.The recommendation method used by the module is fit for not only CRBT, but also wireless music and MRBT (Multimedia Ring Back Tone). It could be helpful in the recommendation process of other products.The first chapter of this paper shows some basic knowledge, including the CRBT service and its development, ETL, association rule theory, Web Services and CPAS. The second chapter of the paper shows a brief introduction of color ring recommendation module, including requirement analysis and brief design description. The third chapter of the paper introduces ETL sub-module. The functions of the ETL sub-module are to synchronize data from data source, and to transform the data to meet the needs. It is the basic of follow-up analysis. The forth chapter of the paper introduces association rule sub-module. The main function of it is to mining multidimensional association rules of CRBT from the color ring order data cube. The fifth chapter of the paper introduces the concept of recommendation weight, and recommendation weight adjust sub-module. The main function of it is to set the recommendation weight of each color ring. The recommendation weight can be set by not only algorithms but also system user. The sixth chapter of the paper introduces recommendation sub-module. The main function of it is to apply recommendation based on multidimensional association rules and recommendation weight. There are many types of recommendation, for example, to recommend ring based on ring, to recommend ring based on user, to recommend user based on ring, etc. The sub-module is able to support offline recommendation and online recommendation.