摘要
Abstract
In the context of the computer information age,traditional library management services cannot meet the diverse read-ing needs of college students.In order to help universities comprehensively understand the needs of readers and provide precise reading services,the study utilizes convolutional and graph convolutional networks to extract features among readers,respec-tively,utilizes self-attention mechanisms to denoise data information,calculates feature weights,and integrates all features to construct a joint reader portrait prediction model based on multi-layer attention networks.The results show that the model di-vides the reader population into four categories,accounting for 45.28％,23.14％,15.09％,and 16.49％,respectively.The learning time of the joint model is 71.06 s,which is 72.23 s and 68.94 s lower than that of the comparative model,respective-ly.The highest accuracy value is 91.09％,and the maximum F1 value is 89.23％,indicating good overall performance.This model can help libraries improve book management and provide precise services.关键词
读者画像/注意力机制/多粒度/日志/高校图书馆Key words
reader portrait/attention mechanism/multi-granularity/log/university library分类
计算机与自动化
