| 英文摘要 |
The traditional file or music recommendation mechanism often requires the user to provide detailed scores for each item. However, drivers will not be able to conduct these tedious procedures while driving. Therefore, special care needs to be taken when designing the feedback procedures for an intelligent car stereo system. In this paper, we use the preference record of users with similar backgrounds to predict a driver's favorite radio stations. The users only need to skip programs they dislike, and do not need to provide detailed feedback on the stations they like. Since radio programs have different schedules, the recommendations also take into account the time factor, which is to say that the system only recommends radio programs that are currently airing or available to the user. In addition, we have also personalized the recommendation mechanism to better meet the needs of different users. For music recommendations, on top of the recommendation mechanism used for radios, we also applied an artificial intelligence solution to learn the user's personal music preferences. It provides recommendations by studying the content of the songs. |