Design And Implementation Of Traffic-oriented Agile Recommended Platform

With the rapid development of mobile Internet, operators accelerate entering into the era of traffic management. In the traditional operator management mechanism, user data and analysis results are scattered in many different systems thus form the "information island" and lead to information missing of customer usage scenario. Customer insight and marketing implementation are based on group rather than individual, therefore the marketing of coarser granularity cannot adapt to the Internet marketing service system of the era. In the face of the diversification of user behavior on the Internet, how to benefit people, how to develop information consumption, and how to control the information security problems etc. are widening. Problems are widespread in using the traffic, such as "dare not to use, do not know how to use, or not good at using it". There is an urgent need to cultivate customer habits, cut off the traffic calmness, and further enhance the convenience and accuracy of the support system for traffic content recommendation.Based on the situation of operators, aiming to forge the ability of sustainable development in traffic management, the dissertation builds up a support system for traffic business aiming at releasing traffic value, in which intelligent pipeline(physical network) and aggregation platform(business network) are the foundation, the directions of management are expanding the scale, promoting the level and enriching the connotation of traffic. The system helps to innovatively develop and provide abundant information application for individual customers, to provide information solutions with significant social benefits for different industries. Thus it promotes the society informatization process and shares information development results.In the process of research and development, several points are finished as follow. First, the current mainstream cutting-edge technologies of big data and cloud computing are researched integratedly, finding out that there are lots of typical application platforms for mobile Internet big data, that technical solutions are intermingled, and many platforms failed to meet the operation requirements of telecommunication in performance and safety level.Then, the technologies related to agile recommending are studied, including stream processing technology and the recommendation algorithm. On the foundation of large data acquisition technology, the project researches and develops out content-based technology of fingerprint depth DPI recognition”, analyzing the functions and protocols within the application, sensing and getting the user information, such as behavior online, location and terminal information. This realizes the integration of new Internet technology and existing marketing support system of operators. By combining the new Internet social network and classic algorithms of traditional expert system, “the adaptive hybrid collaborative filtering recommendation algorithm based on social network model ” is researched and developed, achieving precision recommendation of personalized content to customers. Through the adoption of stream processing technology in the innovative architecture of Active MQ + Kafka + Spark Streaming, “a content recommendation system based on message adaptation ” is researched and developed. This achieves high-speed communication between conventional business message queue and real-time flow engine by message adaptation, and finished the efficient push of recommended contents.Tests show that the system operates and meets the functional requirements, so the project reaches the expected goal. The development of stream processing technology and its industrialized application conform to the urgent needs in current domestic market of large data, and are in line with the guidance of national policy and support. Moreover, this has certain reference value to the key technology development and industrialized application of domestic marketing of mobile real-time transmission.