I. Background
With the rapid development and popularization of Internet technology, the rapid growth in the number of online movies, in order to choose a movie you want to see become increasingly difficult from a number of films. In order to get a better viewing experience, recommendation system came into being.
The system is the solution recommended tool to obtain vast amounts of information in the user wants the data, giving users a good experience.
Second, the project profiles
the work is MovieLens data sets and TMDB site data as the basis, the movie recommendation system based Spark big data platform to build. It contains a recommendation offline and real-time recommendations system. It provides a closed-loop multi-faceted business application front-end, back-end services, from algorithm design and implementation, deployment platform to achieve.
Third, the architecture and data flow diagram
IV describes the data source and processing
five, modular design
1. Offline recommendation module
offline recommendation algorithm module core processes
2. Real-time recommendation module
in real-time recommendation algorithm module Core: Recommended priority is calculated using
3 similar recommendation module
4. Statistical recommendation module
VI deal with the problem of cold start
