基于多头注意力机制下的关键词预测模型——以南京大学图书馆为例  

Analysis of Book Borrowing Hotspots Based on Keyword Prediction Model of Multi Head Attention Mechanism:a Case Study of Nanjing University Library

作　　者：柏政含 何欣楠 徐映千 BAI Zhenghan;HE Xinnan;XU Yingqian(School of Information Management,Nanjing University,Nanjing 210023,China;College of Microelectronics and Communication Engineering,Chongqing University,Chongqing 400044,China;Business School,Hohai University,Nanjing 210024,China)

机构地区：[1]南京大学信息管理学院,江苏南京210023 [2]重庆大学微电子与通信工程学院,重庆400044 [3]河海大学商学院,江苏南京210024

出　　处：《信息与电脑》2022年第22期202-205,222,共5页Information & Computer

摘　　要：在智慧图书馆的研究领域中,通过对读者借阅热点的研究,可以优化高校图书馆的采购工作,提升读者服务质量。文章使用了南京大学图书馆2017—2021年度的借阅数据集,通过拆分借阅书籍名称的关键词,提出了一种基于多头注意力机制模型,以预测借阅关键词的变化趋势。该方法能够准确捕获字符间潜在的依赖权重、语境和语义关联等多方面的特征,以提升预测模型的精确性,同时使用决定系数值作为评判标准,将该模型与传统的多层感知机模型进行可行性和有效性的验证,并对比预测结果。实验表明,利用本文方法使用书籍借阅记录进行关键词的热点预测是可行的,与传统模型相比效果提升显著,具有一定的推广价值。In the research field of smart Library,through the research on reader borrowing hotspots,we can optimize the procurement work of university libraries and improve the quality of reader service.Using the borrowing data set of Nanjing University library from 2017 to 2021,this paper proposes a model based on multi head attention mechanism to predict the change trend of borrowing keywords by splitting the keywords of borrowed book names.This method can accurately capture the potential dependency weight,context,semantic association and other characteristics between characters to improve the accuracy of the prediction model.Using the value of the decision coefficient as the evaluation standard,the feasibility and effectiveness of the model are verified with the traditional multi-layer perceptron model,and the prediction results are compared.Experiments show that the method used in this paper,using book borrowing records to predict the hot spots of keywords,is feasible,and compared with the traditional model in the past,it has significantly improved the effect,and has a certain promotion value.

关 键 词：智慧图书馆 多头注意力机制 VITERBI算法 借阅热点 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术]