比较分析近年来三种图书分类研究模型：基于集成学习的分类模型、基于双向LSTM的分类模型和基于向量空

模型的分类模型。在相同数据集上测试，后两者分类模型的准确率相近，它们相较于第一种模型的准确率高出约2%~8%，其

中基于向量空模型的分类模型用时短；而基于双向LSTM的分类模型的准确率有提升空间，会随训练次数的增加而逐步上升，

但所耗时间也会增加。这几种模型对后期智慧图书馆图书分类起着不可忽视的启示作用，学者们可以根据需要选择使用上述

模型，对于不同分类算法的研究，有机会对分类的准确度、速度实现突破。

This paper compares and analyzes three research models of book classification in recent years, which are the classification

model based on ensemble learning, the classification model based on bidirectional LSTM and the classification model based on vector space model. Tested on the same data set, the accuracy of the latter two classification models is similar, and their accuracy is about 2% ~ 7% higher than that of the first model, in which the classification model based on vector space model takes less time.

The accuracy of the classification model based on bidirectional LSTM can be improved, which will gradually increase with the in

creasing of training times, but the time will also increase. These models play an important role in the book classification of the later

Smart Library. Scholars can choose to use the above models according to their needs. For the research of different classification algorithms, they have the opportunity to make a breakthrough in the accuracy and speed of classification.