Information popularization method and device, electronic equipment and computer readable medium

Abstract

The invention discloses an information popularization method and device, electronic equipment and a computer readable medium. The method comprises the steps that request information is acquired in real time, wherein the request information comprises user information and popularization information; the request information is mapped into a natural language to generate first data; the first data is input into a ranking model to determine ranking scores, wherein the ranking model is a deep convolutional neural network model; and the popularization information in the request information is popularized according to the ranking scores. Through the information popularization method and device, the electronic equipment and the computer readable medium, feature selection, modeling and other work canbe reduced, a large amount of feature engineering is avoided, the generalization ability of a ranking method is enhanced, all features are processed by the adoption of the same network, and model complexity is lowered.

Description

technical field [0001] The present invention relates to the field of big data information processing, in particular to a method, device, electronic equipment and computer-readable medium for information promotion. Background technique [0002] The sorting method is information promotion, such as a core technology in the recommendation / advertising system. The sorting method directly affects the display order of recommended items or advertisements, which in turn affects the user's click and subsequent purchase conversion. A good sorting method can make recommendations / advertisements with better quality rank higher, thereby increasing user clicks and consumption. Therefore, the ranking algorithm is a very critical part of the recommendation / advertising system. At present, the ranking scheme adopted in the recommendation / advertising system mainly includes two steps or modules: first, the extraction and processing of relevant features; and second, the ranking of candidate recomm...

AI Technical Summary

Problems solved by technology

This also leads to two disadvantages: First, a lot of additional feature engineering is required, such as feature screening, determining how to combine features, etc.; second, feature combination can make the sorting method learn a good memory ability, which can be well processed The features and combinations that appear frequently in the training data will also lead to insufficient generalization ability of the sorting method, and the processing of new features and combinations is not good enough

Images