Disney has recently launched a new streaming platform, Disney +. Here, we have a patent application (US2019/0340659A1) by Disney that relates to reviewing of content (similar to how Netflix allows users to review watched content).
The application describes that platforms such as YELP, ROTTEN TOMATOES, AMAZON, TRIP ADVISOR, and NETFLIX allows users to submit reviews differently, for example, stars, points, happy faces, tomatoes, etc. The patent application points out that, although Netflix attempts to learn a user’s overall preference and recommend products or services, those ranking systems are generally binary or numeric. Presumably, features that are discussed in this patent application are somehow non-binary or numeric, although this was not clear upon my review.
A user-specific evaluation profile can be generated based on initial inputs from the user as to preferences for movies, restaurants, travel services, books, consumer products, video games, or personal services of the user. It appears that Disney can gather information about a user’s preferences based on reviews from other platforms (e.g., YELP, ROTTEN TOMATOES, AMAZON, etc.) to initially ‘train’ a profile of the user, although that is not completely clear. The user’s profile is trained based on data of other reviewers with similar reviews, characteristics, etc.
This trained profile can be used to guess how well the user will rate different types of content. Machine learning (e.g., a neural network, decision tree, or linear regression model) will then further hone or ‘train’ the user’s profile to better predict how the user will rate additional content, based on discrepancies between the user ratings and previous predictions.
Presumably, all the big players including Netflix already leverage machine learning to predict what content a user will have a preference towards. Thus, it is unclear what the ‘Gee Whiz’ of this patent application is, but this tends to become more clear during the patent prosecution process.