Research Of Weighted Hybrid Recommendation Technology Based On Multi-objective Optimization Genetic Algorithm

With the rapid development of the Internet, the resources on the internet spring up like the mushrooms and have shown an explosive growth trend. A wealth of information brings us convenience, but it also makes precise information finding more and more difficult. We have officially entered the "information overload" age, in this context, the recommendation system as an effective method to solve this problem have been proposed and attracted more and more attention.Recommendation system as a new expert system, by analyzing the historical behavior of users, helping users find things which may be of interest, and feeding the recommend results back to the users after a certain selection and sorting, is being introduced to a wider range of fields. Although the recommendation algorithm has been applied to many areas, but its development is not perfect, there are still some problems in practical application such as low accuracy, cold start, sparse data and poor real-time property. In this case, a single recommendation algorithm has become out of date, and the hybrid recommendation which can combine different methods has become an inevitable trend.In order to solve the problem of low recommendation accuracy effectively, we further research a new hybrid recommendation algorithm on the base of several common algorithms. Our major contributions are as follows:1. After researching and summarizing the existing hybrid recommendation algorithms, we selecte the weighted hybrid recommendation algorithm as the research center.2. In order to determine the weights of weighted hybrid recommendation and optimize multiple indicators, a multi-objective optimization genetic algorithm is introduced, then a new weighted hybrid recommendation is designed.3. Our experiment is on the Movie Lens dataset. The result shows that the new algorithm's precision and recall have been enhanced compared with several basic collaborative filtering algorithms.4. Aimed at the problem of IT book selection, a simple IT book recommendation system has been built on the LAMP environment. User satisfaction surveys prove this recommendation system can provide effective help when one selects IT books.5. To solve the problems that simple IT books recommendation systems face in the practical application, some improvement schemes are proposed.