Mobile APP popularity prediction model based on CNN-GRU
Mobile app popularity prediction is crucial for app recommendation and advertising.However,existing methods rely heavily on manual feature engineering,which is labor-intensive and inefficient.This paper proposes a deep neural net-work model for mobile app popularity prediction.First,the maximal information coefficient is utilized to perform feature cor-relation analysis,ensuring the effectiveness of feature selection.Combined with historical popularity features,a long-term e-volution model is constructed through a gate recurrent unit(GRU)and an attention mechanism to deduce the development trend.A short-term fluctuation model is constructed based on multi-scale convolutional neural networks(CNN)and an at-tention mechanism to achieve dynamic optimization of prediction.Also,other important features are incorporated to build multi-factor influence models using GRU and attention.Finally,the above models are integrated through a temporal atten-tion module to achieve popularity prediction.Experimental results demonstrate the relative accuracy and effectiveness of the proposed model in predicting the popularity of mobile apps.
mobile apppopularity predictionattention mechanismconvolutional neural networks(CNN)gate recurrent unit(GRU)
国家科技期刊平台
NSTL
万方数据
