Context sensing recommendation method based on deep neural network

Abstract

The invention provides a context sensing recommendation method based on a deep neural network, and belongs to the technical field of Internet information recommendation. The recommendation method is constructed by using a user feature vector and a recommendation object feature vector, and text features of each recommendation object are extracted by using a convolutional neural network algorithm; constructing a text embedding vector for each recommendation object, and extracting a potential feature vector of a user by using a matrix decomposition technology; the method does not need professional domain knowledge and individual information, and is safe and simple; the minimum root-mean-square error is adopted as an optimization constraint condition, a correct prediction score can still be given only by training a score matrix in the implementation process, and the problem of prediction errors caused by data sparsity is avoided; according to the method, a recommendation user potential feature matrix and a recommendation object embedding matrix are fused by using a deep neural network to obtain a high-dimensional recommendation object feature vector, and joint training is performed, so that the recommendation method is prevented from being over-fitted, and a recommendation result is optimized.

Description

technical field [0001] The invention belongs to the technical field of Internet information recommendation, and in particular relates to a context-aware recommendation method based on a deep neural network. Background technique [0002] With the rapid development of the Internet and the rapid popularization of electronic application products, various fields in a diversified society are flooded with a large amount of information. One of the effective solutions to the problem of information overload is a personalized recommendation system. The recommendation system discovers the user's points of interest, thereby guiding users to discover their own information needs. Personalized recommendation systems are widely used in many fields, especially in the field of e-commerce. In academia, recommender systems have gradually become an independent discipline. The recommendation method is the core and key part of the recommendation system, which largely determines the type and perfo...

AI Technical Summary

Problems solved by technology

The main disadvantage is that although there are many recommended combination methods in theory, the scalability of this system is not good, and there is no standardized and exact method for solving the problem;

[0016] As an effective means of information filtering, the personalized recommendation system is one of the effective methods to solve the problem of information overload and realize personalized information services; however, with the explosive growth of the number of users and the number of services, the high dynamic And the continuous complexity of the data generated by the service makes the service-oriented recommendation system have to face some new challenges, especially the high-dimensional problem of data characteristics, the problem of data coefficient and the problem of cold start

Images