A Method for Movie Recommendation Using User Attributes

Abstract

The invention discloses a method for recommending movies by using user attributes, which belongs to the field of data recommendation, and aims to solve the problem that traditional recommendation model methods use auxiliary information unreasonably, resulting in inaccurate recommendation results and even large deviations. The traditional recommendation system model does not discriminate some attribute information of the user, and ignores the adaptability of the attribute to the corresponding item, so there is a large irrationality. This method adds attention to the attribute utilization link in the traditional recommendation system workflow. Mechanism, so as to effectively control and decide which user information is kept and which is discarded when recommending movies, making the whole recommendation process more reasonable and effective. At the same time, this method combines large-scale parameter learning of deep learning, which makes the model more applicable and robust.

Description

technical field [0001] The invention relates to the field of data recommendation, in particular to the prediction rating and recommendation of movies, and in particular to a method for movie recommendation using user attributes. Background technique [0002] In the era of big data, only by making full use of data can we take the lead in business competition. The same is true for the movie recommendation system. Only by allowing the system to fully learn the historical movie viewing data of a large number of users, select movies that users may like from the movie library, and make accurate recommendations can the distribution of movies be more effectively promoted. However, most movie recommendation systems face such a cold start problem: for new users, there is little or no movie viewing history data, so it is difficult to recommend them. Based on the above problems, some methods have begun to model user attributes, such as age, gender, and occupation, which will be entered...

AI Technical Summary

Problems solved by technology

However, most movie recommendation systems face such a cold start problem: for new users, there is little or no movie viewing history data, and it is difficult to recommend them

These methods take into account the modeling of user attributes and aim to solve the recommendation problem for new users, but they are more or less irrational

[0004] There is such a problem in the above-mentioned models for modeling user attributes: no matter which movie is recommended for the target user, when the system models the user model, the proportion of the same attribute in the feature space is the same

Consider a practical problem: for the recommendation of action movies, the system should pay more attention to the gender of the user, and for the recommendation of romantic love movies, it should pay attention to the age of the user, instead of using the same attributes for all movies, this will large forecast bias

Images