Design of Distributed Recommendation Engine Based on Hadoop and Mahout
Jian Yun Yu · Applied Mechanics and Materials · 2014
The distributed recommendation engine consists of three layers of data storage layer, Produce recommended layer and application layer, the data storage layer is mainly stored user preferences data, these data are recommended on the basis of upper recommendation engines. Produce recommend layer producing part recommend the key lies in the recommendation engine of the algorithm, the algorithm adopts the Mahout as recommendation framework, and implement custom recommendation algorithm, including the recommendation algorithm based on user similarity, based on the recommendations from the project similarity algorithm and based on the recommendations of the Slope One algorithm after receiving recommended the client's request, the Servlet will produce a recommended by pushing engine first get data model, according to the similarity between data model computing project, and generate the recommended ID. The application layer is mainly based on B/S architecture to implement, can be very easily expanded to mobile platforms.