构建经济文本情感分析专用词典，提升领域内情感判断准确率。
Sentiment Analysis of Economic Text: A Lexicon-Based Approach
- 针对经济学文本设计专用词典，覆盖广泛经济术语
- 人工标注情感得分范围为[-1,1]，更精准反映语义倾向
- 相比其他词典，覆盖更全、情感分类更准，适合经济研究
我们提出一种专为经济文本应用设计的经济词典（EL），具备两大特点：一是广泛覆盖经济概念相关术语，二是提供[-1,1]范围内的人工标注情感得分。通过简单情感度量的应用示例，展示了该词典在经济学中的多种用途。与现有词典对比表明，该词典因涵盖更多领域相关词汇且情感分类更准确而表现更优。
原文摘要 · Abstract (English)
We propose an Economic Lexicon (EL) specifically designed for textual applications in economics. We construct the dictionary with two important characteristics: 1) to have a wide coverage of terms used in documents discussing economic concepts, and 2) to provide a human-annotated sentiment score in the range [-1,1]. We illustrate the use of the EL in the context of a simple sentiment measure and consider several applications in economics. The comparison to other lexicons shows that the EL is superior due to its wider coverage of domain relevant terms and its more accurate categorization of the word sentiment.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。