Online and offline hybrid recommendation method and system

Abstract

According to an online and offline hybrid recommendation method and system provided by the invention, the precision is improved and the problem of cold start is solved by carrying out deep learning onthe total data. And meanwhile, the offline use condition is corrected in real time according to the use condition of each user, so that the problems of individuation and response speed are solved. Compared with a traditional scheme, the online and offline mixed scheme uses an offline model as an intermediate result of the whole recommendation service. The intermediate result is more excellent inperformance due to the fact that full data is used for learning. The method is mainly used for solving the problems of cold start and data scarcity. And the online model is a lighter model, and the result of the offline model is combined with the preference of the current enterprise user to generate a result. By combining the offline model and the online model, the personalized demands of different enterprises are ensured while the recommendation quality is considered, and the user experience is improved; and through an automatic training pipeline, the weight is basically, fully-automaticallyand automatically updated, so that the maintenance cost of the model is reduced.

Description

technical field [0001] The invention relates to the field of SaaS platform recommendation systems, in particular to an online and offline hybrid recommendation method and system. Background technique [0002] SaaS is the abbreviation of Software-as-a-Service (Software as a Service). With the development of Internet technology and the maturity of application software, a completely innovative software application model began to emerge in the 21st century. It has a similar meaning to "on-demand software" (on-demand software), the application service provider (ASP, application service provider), and hosted software (hosted software). It is a mode of providing software through the Internet. The manufacturer uniformly deploys the application software on its own server. Customers can order the required application software services from the manufacturer through the Internet according to their actual needs, according to the number of services ordered and the length of time. Pay the...

AI Technical Summary

Problems solved by technology

[0009] (2) Lack of personalization: Although many recommendation systems have introduced the concept of thousands of people and faces, they mainly cluster and recommend users, and do not really customize preferences for different users;

[0010] (3) The learning speed is not fast enough: the update frequency of the model based on the deep learning model algorithm is usually slower, and the process of learning the data offline and then updating the model weights is required

Images