曹莎莎 徐岚 刘涓

摘 要 随着读者需求驱动采购在高校图书馆的应用日益广泛，随之而来的经费超支、馆藏结构失衡等问题引起业界关注，为有效解决上述问题，文章通过探索机器学习方法在读者需求驱动采购中的应用，构建具体应用框架，并从读者、图书、模型三个角度对应用框架进行拓展和延伸，分别描述不同角度下机器学习方法应用于读者需求驱动采购的技术路线。研究表明将机器学习方法应用于读者需求驱动采购，可以有效预测读者需求、图书触发采购概率和馆藏结构，从而实现节省图书馆经费、改善馆藏结构失衡的目标，事实证明其对图书馆采购决策和馆藏建设有着积极的影响。关键词 机器学习  读者需求驱动采购  读者决策采购  馆藏建设

分类号 G253.1

DOI 10.16810/j.cnki.1672-514X.2021.07.012

Analysis on the Application of Machine Learning in Demand Driven Acquisition

Cao Shasha，  Xu Lan，  Liu Juan

Abstract With the wide application of demand driven acquisition in university libraries， the following problems such as over expenditure and unbalanced collection structure have attracted the attention of libraries. In order to solve the above problems effectively， the paper first puts forward a specific application framework by exploring the application of machine learning in demand driven acquisition， then expands and extends the application framework from the perspectives of readers， books and models， and respectively describes the technical route of the application of machine learning methods in demand driven acquisition from different angles. The research shows that the application of machine learning method in demand driven acquisition can effectively predict readers demand， the likelihood of books being triggered for purchase and collection structure， so as to achieve the goal of saving library funds and improving the imbalance of collection structure. Facts have proved that it has a positive impact on library procurement decision-making and collection construction.

KeywordsMachine learning. Demand driven acquisition. Patron driven acquisition. Collection development.

0 引言

图书馆一直有读者参与图书采购决策的传统，从读者意见箱到线上荐购再到馆际互借，但馆藏建设的最终决策权仍在图书馆员身上。随着读者对文献资源需求越来越高，电子出版物大量涌现，与此同时，图书馆经费收缩，现有馆藏利用率较低，催生了读者需求驱动的图书采购新模式[1]（Demand Driven Acquisition，以下简称DDA）。ProQuest2018年发布题为“Why DDA is Here to Stay”的白皮书[2]，对全世界449名图书馆员（其中99%的受访者为高校图书馆员）开展调查，有93%的受访者称其图书馆采用多种电子书采购模式，92%的受访者称需求驱动采购是其图书馆电子书的主要采购模式，可见读者决策采购在国外图书馆，特别是高校图书馆扮演着越来越重要的角色，学界一致认为该采购方法是对当前馆藏建设方法的有益補充[3]。……