Culturomics is emerging as an important field within science, a way to measure attitudes and beliefs and their dynamics across time and space via quantitative analysis of digitized data from literature, news, film, social media, and more. Sentiment analysis is an emerging tool for culturomics that, within the last decade, has provided a means to quantify the polarity of attitudes expressed within various media. Conservation science is a crisis discipline and one in which accurate and effective communication are paramount to success. We investigated how conservation scientists communicate their findings through one of their primary media: scientific journal articles. We analysed 15,001 abstracts from papers published in conservation-focused journals published in the last 20 years, 1998-2017. Papers were categorized by year, focal taxa, and their conservation status; the mean sentiment scores were then extracted from the abstract using four lexicons (Jockers-Rinker, NRC, Bing et al., and AFINN). We found no annual trend in the sentiment scores of papers across conservation literature but analysis of absolute values suggested increasing polarization of language over time (i.e. less neutral). We also observed a trend towards increasing negativity along the spectrum of IUCN Red List categories (i.e. from Least Concern to Critically Endangered to Extinct), though this relationship was not significant. There were some clear differences in the sentiments with which research on different taxa were reported, however. For example, abstracts mentioning lobe finned fishes tended to have high sentiment scores, which we hypothesize may be related to the rediscovery of the coelacanth driving a positive narrative. Contrastingly, abstracts mentioning elasmobranchs had low scores, reflecting the negative sentiment score associated with the word "sharks". Sentiment analysis is an exciting frontier with applications in science and we suggest a new science-based lexicon be developed for applying this tool to conservation. This article is protected by copyright. All rights reserved.

中文翻译：

---

情感分析作为科学文献中保护文化的衡量标准

文化组学正在成为科学中的一个重要领域，它是一种通过对来自文学、新闻、电影、社交媒体等的数字化数据进行定量分析来衡量态度和信仰及其跨时空动态的方法。情感分析是文化组学的新兴工具，在过去十年中，它提供了一种量化各种媒体中表达的态度极性的方法。保护科学是一门危机学科，准确有效的沟通对成功至关重要。我们调查了保护科学家如何通过他们的主要媒体之一：科学期刊文章来传达他们的发现。我们分析了过去 20 年（1998 年至 2017 年）发表在以保护为重点的期刊上的论文的 15,001 篇摘要。论文按年份、焦点分类群、及其保护状况；然后使用四个词典（Jockers-Rinker、NRC、Bing 等人和 AFINN）从摘要中提取平均情感分数。我们发现保护文献中论文的情绪分数没有年度趋势，但对绝对值的分析表明，随着时间的推移，语言的两极分化越来越大（即不那么中立）。我们还观察到在 IUCN 红色名录类别（即从最不关心到极危到灭绝）的范围内消极性增加的趋势，尽管这种关系并不显着。然而，对不同分类群的研究报告的情绪存在一些明显的差异。例如，提到叶鳍鱼类的摘要往往具有较高的情感得分，我们假设这可能与重新发现腔棘鱼有关，推动了积极的叙述。相比之下，提到 elasmobranchs 的摘要得分较低，反映了与“鲨鱼”一词相关的负面情绪得分。情感分析是一个在科学中应用的激动人心的前沿领域，我们建议开发一种新的基于科学的词典，以将该工具应用于保护。本文受版权保护。版权所有。本文受版权保护。版权所有。本文受版权保护。版权所有。