预训练模型在“你选书,我买单”计算机系统中的应用

Application of pre-training model in computer system of“You choose the book,I pay the bill”

为了解决图书馆“你选书,我买单”计算机系统购书审核存在的问题,文章提出引入人工智能深度学习方法,在原有计算机系统中加入购书审核模型。文章基于预训练模型BERTbase模型、BERT-wwm模型和Xlnet模型,构建“你选书,我买单”购书审核模型,微调训练这3个模型,计算3个模型验证结果,并对3个模型进行了比较。结果表明,BERTbase模型、BERT-wwm模型和Xlnet模型验证结果Accuracy、Precision、Recall和F1值达到设计要求,其中BERT-wwm模型和Xlnet模型各项评价指标都高于BERTbase模型。

In order to solve the problems existing in the book purchase audit of the library’s“You choose books,I pay the bill”computer system,this paper puts forward the introduction of artificial intelligence deep learning method,and adds the book purchase audit model to the original computer system.Based on the pre-training model BERTbase model,BERT-wwm model and Xlnet model,the“You choose books,I pay the bill”book purchase audit model was constructed,and the three models were fine-tuned for training,and the verification results of the three models were calculated and compared.The accuracy,precision,recall and F1 values of BERTbase model,BERT-wwm model and Xlnet model meet the design requirements,and the evaluation indexes of BERT-wwm model and Xlnet model are higher than BERTbase model.