学术应用

一种用于发现偏见(刻板印象)的交互式可视化工具An interactive visualization tool for spotting biases (stereotypes)...

为推动我国经济、统计等社会科学量化研究方法学习与应用，培养和训练社会科学相关领域的青年学者、硕博士研究生，促进社会科学相关领域研究方法科学化规范化，“结构模型、DSGE、Stata实证前沿、空间计量、Python数据挖掘”五一工作坊为广大学者提供了一个高水平学术交流、研究方法普及与研究经验分享的平台。工作坊采用模块式教学方法，不仅侧重经济、统计等社会科学量化基本方法的介绍，而且更加注重研究设计与研究选题训练，注重理论实践相结合，培养学员社会科学量化分析研究的综合能力。

人类在留下语言、文字的过程中，也留下了自己的偏见、态度等主观认知信息（偏见、态度）。通过词向量距离的测算，就可以间接测得不同群体 对 某概念(组织、群体、品牌、地域等)的态度偏见。...

语言具体性描述了一个词在多大程度上是指一个实际的、有形的或“真实的”实体，以一种更具体、更熟悉、更容易被眼睛或心灵感知的方式描述对象和行为（即，可想象或生动；Brysbaert, Warriner, and Kuperman 2014; Semin and Fiedler 1988). 我找了三篇论文，简单分享一下。...

在大数据时代的背景下，基于大数据的分析处理技术为以“数据驱动”的社会科学研究创造了新的发 展契机。其中，词嵌入(Word Embeddings)技术借势大数据浪潮，以其高效的词表征能力和强大的迁移学习 能力在文本分析领域受到越来越多的关注。不同于传统的文本分析路径，词嵌入技术不仅实现了对非结构 化文本数据的表征，还保留了丰富的语义信息，可以实现对跨时间、跨文化文本中深层次文化信息的挖掘， 极大丰富了传统的社会科学实证的研究方法。文章总结了词嵌入技术的基本原理及特点，系统地梳理了词 嵌入技术的六大应用主题：社会偏见、概念联想、语义演变、组织关系、文本情感和个体决策机制。随后， 文章归纳了词嵌入技术的基本应用流程。词嵌入技术还面临文本数据的选择、中文文本的分词处理、单词 语义信息的表征层次三种挑战，文章归纳了相应的应对思路与方法。最后，基于词嵌入技术的强大适应能 力，未来研究可以进一步关注该技术在管理领域的应用前景，包括政策效应评估、用户推荐系统、品牌管 理、企业关系管理、组织内部管理、中国传统智慧与管理问题六个方面。In the context of the era of big data, the analysis and processing technology based on big data has created new development opportunities for data-driven social science research. Among them, word embedding (Word Embeddings) technology takes advantage of the wave of big data, and has received more and more attention in the field of text analysis with its efficient word representation ability and powerful transfer learning ability. Different from the traditional text analysis path, the word embedding technology not only realizes the representation of unstructured text data, but also retains rich semantic information, which can realize the mining of deep cultural information in cross-time and cross-cultural texts, which greatly It enriches the traditional social science empirical research methods. This article summarizes the basic principles and characteristics of word embedding technology, and systematically sorts out six application themes of word embedding technology: social bias, concept association, semantic evolution, organizational relationship, text emotion, and individual decision-making mechanism. Subsequently, the article summarizes the basic application process of word embedding technology. Word embedding technology also faces three challenges: the selection of text data, the word segmentation processing of Chinese text, and the representation level of word semantic information. The article summarizes the corresponding countermeasures and methods. Finally, based on the strong adaptability of word embedding technology, future research can further focus on the application prospects of this technology in the field of management, including policy effect evaluation, user recommendation system, brand management, enterprise relationship management, organization internal management, traditional Chinese wisdom and management There are six aspects to the problem....