图书馆借阅系统的升级与创新是提升图书馆服务质量和读者体验的关键,也是智慧图书馆建设的重要工作.本研究通过采集图书馆的借阅信息、读者信息和图书信息等数据,采用基于Transformer的双向编码(Bidirectional Encoder Representations from Transformers,BERT)模型提取图书特征,应用多层感知机(Multilayer Perceptron,MLP)深度学习方法,对读者的历史借阅记录信息进行全面的数据挖掘,分析读者的借阅偏好.结果表明,BERT-MLP模型的性能明显优于基础神经网络模型,且可以更有效地找到图书推荐数据的重要特征.本研究可为提高图书馆个性化服务水平提供理论依据.
Research on Book Resource Borrowing Recommendation AlgorithmBased on Deep Learning
The upgrading and innovation of library borrowing system is the key to improve library service quality and reader experience,and is also an important work in the construction of smart library.By collecting data such as library borrowing information,reader information and book information,this study adopts Bidirectional Encoder Representations from Transformers(BERT)model to extract book features,and applies Multilayer Perceptron(MLP)deep learning method to carry out comprehensive data mining on readers'historical borrowing record information and analyze readers'borrowing preferences.The results show that the BERT-MLP model performs better than the basic neural network model and can find the important features of book recommendation data more effectively.This study can provide theoretical basis for improving the level of personalized library service.
deep learningMultilayer Perceptron(MLP)Bidirectional Encoder Representations from Transformers(BERT)recommendation algorithm
NETL
NSTL
万方数据
