An object recommendation system, method, apparatus, electronic device and storage medium

Abstract

The present disclosure relates to an object recommendation system, method, device, electronic equipment, and storage medium, and belongs to the field of Internet technology. In the system, multiple storage devices are used to store the sparse features of each account and the embedding data corresponding to the sparse features of each account. parameters, where the sparse features of each account are generated based on the account’s operational behavior data, and the embedding parameters are pre-learned by the prediction model; the reasoning server is used to determine when the conditions for recommending objects to any target account are satisfied, Obtain the sparse features of the target account from the storage device corresponding to the target account, determine the storage device corresponding to the sparse feature according to the preset storage method of the embedded parameters, read the embedded parameters corresponding to the sparse feature from the determined storage device, and then use each sparse The embedding parameters corresponding to the features and the stored network parameters of the prediction model are used to determine the recommendation object of the target account, wherein the network parameters are parameters learned by the prediction model other than the embedding parameters.

Description

technical field [0001] The present disclosure relates to the technical field of the Internet, and in particular to an object recommendation system, method, device, electronic equipment and storage medium. Background technique [0002] In the field of Internet technology, each application has enabled a large-scale object recommendation system in order to better retain users. Each reasoning server in the object recommendation system has strict requirements on the scalability of the underlying storage system in terms of capacity, delay, and throughput. [0003] Generally, the model parameters (including network parameters and embedding parameters) of the predictive model used by the object recommendation system all require terabyte (TB) level storage space. In related technologies, in order to share these model parameters among multiple inference servers in an object recommendation system, a collaborative deployment scheme is proposed. figure 1 is a storage schematic diagram ...

AI Technical Summary

Problems solved by technology

[0004] The present disclosure provides an object recommendation system, method, device, electronic equipment and storage medium, to at least solve the problem of relatively high hardware cost of the object recommendation system in the related art

Images