Recommendation for team projects: an automatic music recommendation system based on social networks -- by zhongxia

It has been nearly two months since I came to Microsoft's Asia Research Institute. Everyone is slowly adapting to this compact research atmosphere. I often hear that the advanced software engineering (ASE) mentioned by my senior brother also started on schedule. After class last week, we completed the team. After preliminary research and consideration, I recommended the "social network-based Automatic music recommendation system" as the topic of our team's project.

Music is an important part of our lives. At work, we may need light music to calm ourselves down and stay focused. After work, we may need soothing music to relax in all directions. When we are happy, we are happy to hear some cheerful music; when we are sad, we may need some sad voices to let ourselves go; when we are low, we want to cheer ourselves up with a passionate voice. It can be seen that the status and habits of a person greatly affect the demand for music, which makes the constant favorite list function of most music players unable to meet users' requirements, as a result, the user may frequently adjust the favorite list, which affects the user experience. Based on this problem, designing a better automatic music recommendation system has become my focus.

Project Analysis: Because offline music is limited, there is limited room for automatic music recommendation on offline devices, and there is limited information available, which is difficult to achieve. Therefore, the goal is to aim online. The individual's music playing habits require learning, and the social network status is closely related to the person's status. Therefore, it is natural to think of the social network information as its training set and prior knowledge, based on this recommended music. The following describes the nabc analysis method.

Need, there is almost no such problem. However, in fact, everyone's liking for music is closely related to their status. Poor Player recommendations will greatly affect users' mood. They often make listening to music a counterproductive activity, and users will feel less comfortable with the player, therefore, it is necessary to solve this problem. Such a playing system has a wide audience and has a clear market demand.

Approach: in general, this project involves the following aspects:

1. capture and analysis of social network information;

2. machine learning is used to accurately classify user statuses;

3. Select a type of music based on the user status and other information (such as whether it is working hours), score the music based on previous user behaviors, and recommend music based on the score;

4. Play recommended music and UI design.

It basically includes all aspects of the content, the workload is sufficient, you can also get training in all aspects, easy division of labor, can meet the needs of the course.

Benefits: saves the user's time and suits the user's taste. It can give the user a good listening experience, help the user to have a good status, and stimulate the greatest benefit of music; on the other hand, the integration with social networks is feasible.

Competition, if both the efficiency and effect are good, you will undoubtedly be welcomed.

To sum up, I think that the "social network-based Automatic music recommendation system" is essential for our team's project questions, so I recommend this question. Of course, the final questions of the Team also need further discussion and research in the group.

　　

 

Recommendation for team projects: an automatic music recommendation system based on social networks -- by zhongxia