arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2110.04040cs.IRcs.AIcs.CL
收藏

面向数学感知的科学出版物自动分类和相似性搜索:数学内容表示方法

Towards Math-Aware Automated Classification and Similarity Search of Scientific Publications: Methods of Mathematical Content Representations

Faculty of Informatics, Masaryk University(马萨里克大学信息学院)

Michal Růžička, Petr Sojka

摘要:

在本文中,我们调查的数学内容表示适用于自动分类和STEM文档中的相似性搜索使用标准的机器学习算法:潜在狄利克雷分配(LDA)和潜在语义索引(LSI)。这些方法在arXiv.org论文的一个子集上进行评估,数学主题分类(MSC)作为参考分类,并使用标准的精确度/召回率/F1度量指标。这些结果让我们深入了解不同的数学表示如何影响STEM存储库中分类和相似性搜索任务的性能。毫不奇怪,机器学习方法能够从文本标记中获取分布式语义。适当选择表示数学的加权标记可以稍微提高结果的质量。一个结构化的数学表示,模仿成功的文本处理技术与数学,产生更好的结果比平面TeX令牌。

英文摘要:

In this paper, we investigate mathematical content representations suitable for the automated classification of and the similarity search in STEM documents using standard machine learning algorithms: the Latent Dirichlet Allocation (LDA) and the Latent Semantic Indexing (LSI). The methods are evaluated on a subset of arXiv.org papers with the Mathematics Subject Classification (MSC) as a reference classification and using the standard precision/recall/F1-measure metrics. The results give insight into how different math representations may influence the performance of the classification and similarity search tasks in STEM repositories. Non-surprisingly, machine learning methods are able to grab distributional semantics from textual tokens. A proper selection of weighted tokens representing math may improve the quality of the results slightly. A structured math representation that imitates successful text-processing techniques with math is shown to yield better results than flat TeX tokens.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑