深度学习在图书馆文本分类中的应用研究进展

A Review of Research on the Application of Deep Learning in Library Text Classification

文本分类是图书馆领域的重要研究方向之一。基于深度学习方法对用户生成的内容进行分类有助于图书馆更精准的了解用户行为并评估图书馆服务质量。通过对图书馆领域和使用图书馆数据的计算机科学领域使用深度学习方法进行文本分类的研究进行批判性审查,调查图书馆领域使用深度学习方法进行文本分类的研究现状,为未来研究提出建议。研究结果表明,目前的研究主要集中在文本特征分类、文本情感分类和文本评级分类上。大多数研究仍采用传统的深度学习方法如前馈神经网络和人工神经网络等。近年来在计算机科学领域提出的具有更好分类性能的深度学习算法尚未引入图书馆领域。研究建议引入深度学习算法的方法框架、构建和开发更深层次的算法、明晰文本分类的详细步骤、明确数据标注规范和标注步骤并积极使用结合各自优势的多模型分类方法。

Text categorization is a major research direction in the field of library.Classifying user-generated content based on deep learning methods can help libraries understand user behavior more accurately and evaluate the quality of library services.This paper makes recommendations for future research by investigating and critically reviewing research on the use of deep learning methods for text classification in the library field and in the computer science field using library data.The results show that the current research mainly focuses on text feature classification,text sentiment classification and text rating classification.Most research still uses traditional deep learning methods such as feedforward neural networks and artificial neural networks.Deep learning algorithms with better classification performance in computer science have not been introduced to the library domain.The study recommends introducing a methodological framework for deep learning algorithms,building and developing deeper algorithms,clarifying detailed steps for text classification,specifying data annotation specifications and annotation steps,and actively using multi-model classification methods that combine their respective strengths.