Scoring determination method and device and electronic equipment

Abstract

The embodiment of the invention provides a score determination method and device and electronic equipment. The scheme is as follows: obtaining target feature data of a target user; depending on targetfeature data, determining a similar user similar to the target user from the plurality of sample users; and, according to a first feature vector of a similar user in a basic domain and a second feature vector of each to-be-scored article included in a trained preset matrix model, determining an expected score of the target user for each to-be-scored article, wherein the preset matrix model comprises a conversion matrix and first and second feature vectors on different data domains, the conversion matrix is used for converting the first feature vectors corresponding to the feature matrixes ofthe different data domains, and the conversion between the second feature vectors corresponding to the feature matrixes of the different data domains is used for recommending the to-be-scored articleto the target user. Through the technical scheme provided by the embodiment of the invention, the accuracy of score determination is effectively improved.

Description

technical field [0001] The present invention relates to the technical field of the Internet, in particular to a scoring determination method, device and electronic equipment. Background technique [0002] With the continuous development of Internet technology, there are more and more items on the Internet, such as videos, e-books, commodities and so on. With the increase of items on the network, how to accurately determine the items that users may be interested in is particularly important. In a common manner, items that the user is interested in can be determined according to the user's rating of each item. [0003] When determining the user's rating for each item, not all users have a corresponding rating for each item, therefore, it is inevitable to determine the user's expected rating for each item. At present, when determining the expected rating, the user feature data of multiple sample users, the item feature data of multiple items to be rated, and the historical ra...

AI Technical Summary

Problems solved by technology

[0004] However, in the training process of the joint matrix model, the characteristic data of the dimension with a small amount of data in the training data is often ignored, and only the characteristic data of the dimension with a large amount of data is considered, which leads to the first eigenvector and The second eigenvector is not accurate enough, which affects the accuracy of scoring determination

Images