Movie recommendation algorithm --- HHR plan

1, look at the FM section.

2, take a look at cold start.

     0, recall popular source.

     1, men and women recall the source, age recall the source, the source recalled occupation, the highest score.

     2, men and women age, occupation combined with each other;

     3, stored in redis. Day level update.

3, ordering the recall + first get to know.

4, a sort of a sleeve inside. (Electricity supplier in accordance with the practice, rmse, auc)

 

------ 1, the data stage ----------

ratings.dat: user_id, movie_id, rating, timestamp

users.dat: user_id, gender, age, occupation(职业), zip-code

movie.dat: movie_id, title, genres(体裁).

 

Guess you like

Origin www.cnblogs.com/yueyebigdata/p/11375126.html