基于LSTM神经网络的馆际互借量预测研究

Research on prediction of interlibrary lending based on LSTM neural network

针对高校出现的跨校区多馆藏布局,文章对比采用LSTM神经网络结构对馆际互借的委托数据进行分析、预处理与建模。结果显示:基于LSTM神经网络模型的委托量预测与典藏调整后,委托量下降明显。相较于按借阅比例分配的委托调度,基于LSTM神经网络模型的预测在时间和分类等方面对委托量影响更加显著,对单本书实现委托预测与分析更加贴合实际情况。

In view of the cross campus multi collection layout in universities,this paper analyzes,preprocesses and models the entrusted data of interlibrary loan by using LSTM neural network structure.The results show that after the prediction of entrustment based on LSTM neural network model and the adjustment of the collection,the entrustment decreases obviously.Compared with the entrustment scheduling according to the borrowing proportion,the prediction of entrustment based on LSTM neural network model has a more significant impact on the entrustment amount in terms of time and classification,and the prediction and analysis of entrustment for a single book is more in line with the actual situation.