Mobile advertisement recommendation method based on cluster

Abstract

The invention discloses a mobile advertisement recommendation method based on cluster. The method comprises the following steps: establishing a mobile advertisement putting platform and collecting user data information through the mobile advertisement putting platform; extracting advertisement behavior data from the user data information of the mobile advertisement putting platform; converting the advertisement behavior data to an advertisement behavior vector matrix, each row vector of the advertisement behavior vector matrix being advertisement behavior data of each user; carrying out dimension-reduction calculation on each row vector of the advertisement behavior vector matrix one by one; carrying out user cluster according to the similarity of each user and calculating user clusters; searching advertisement types matched with the user clusters on the mobile advertisement putting platform according to the features of the user clusters and adding the advertisement types to advertisement recommendation sequences of the users; and recommending the mobile advertisements in the advertisement recommendation sequences to the users to put. The mobile advertisement recommendation method based on cluster has the advantage of high accuracy in customized advertisement recommendation.

Description

technical field [0001] The invention relates to the field of communication technology, in particular to a clustering-based mobile advertisement recommendation method. Background technique [0002] Personalized advertising is to recommend relevant advertisements to users based on their browsing interests. It is a very important means of precise advertising. Since the recommended advertisements are closely related to the interests of users, personalized advertising can effectively improve user experience. To make it easier for users to accept, from the perspective of advertisers, personalized advertisements are targeted, avoiding the waste of advertising and improving the efficiency of advertising. [0003] With the optimization of the mobile network access environment, the reduction of network access fees and the popularization of intelligent mobile communication terminals (referred to as mobile terminals), mobile network communication has entered a period of vigorous develop...

AI Technical Summary

Problems solved by technology

[0006] However, when the click-through rate of the advertisement is very low, for example, the click-through rate of the advertisement is between 1 / 10,000 and 1 / 1,000, it is difficult to achieve better and stable results by training with existing historical data. When using historical data for training, there will be a problem of data sparsity, resulting in mixed effects, and most of the clicks on ads may only be 0 or 1, so it is difficult to accurately explain the problem with such a small number of clicks. The probability of clicking is very low

Therefore, the existing method that only relies on the click-through rate of advertisements as an advertisement recommendation element cannot really find out the advertisement delivery target, the accuracy of advertisement recommendation is low, and the stability of the recommendation results is poor.

Images