Detecting the underlying human values within arguments is essential across various domains, ranging from social sciences to recent computational approaches. Identifying these values remains a significant challenge due to their vast numbers and implicit usage in discourse. This study explores the potential of emotion analysis as a key feature in improving the detection of human values and information extraction from this field. It aims to gain insights into human behavior by applying intensive analyses of different levels of human values. Jafari, Amir & Rajapaksha, Praboda & Farahbakhsh, Reza & Li, Guanlin & Crespi, Noel. (2024). Unveiling Human Values: Analyzing Emotions behind Arguments. Entropy. 26. 327. 10.3390/e26040327.
Transitioning to level 2 of our human values analysis, we utilized the original configuration of the emotion model with 27 emotions to obtain a broader perspective of various emotion distributions. Our analysis focused on fine-grained emotions, consisting of 27 emotions categorized into three main groups: positive, negative, and ambiguous. As depicted in Figure 4, in positive emotions, most human values align with the Approval category followed by optimism, admiration, and caring as the next favored positive emotions. The other positive emotions contain a small portion of overall distributions. Considering negative emotions, disapproval and annoyance emerge as dominant categories. This observation suggests that during arguments centered on negative opinions, individuals often aim to express disapproval of certain values or experiences that evoke annoyance, while positive opinions tend to convey approval of specific values. Furthermore, the prevalence of realization among ambiguous
id: 19d2cd6568e385b42dba993eeb57e68d - page: 6
Notably, these emotional nuances hold significant potential as additional features for enhancing the classification process.
id: 37e4997e0a9af74a745bf8f8a93db00f - page: 6
Moreover, given that the dataset comprises arguments structured with a premise, a conclusion, and a stance indicator specifying whether the premise supports or opposes the conclusion, it is noteworthy that the extracted emotions predominantly reflect opposing sentiments, such as approval and disapproval. In conclusion, the extracted information from emotion distribution in all human value levels highlights the significance of particular emotions with varying polarities in human value arguments, suggesting their potential as valuable features for enhancing the performance of human value detection algorithms which are presented in the next section.
id: a7640b3849dbcf3cd9d5be2b14b82326 - page: 6
Figure 4. Fine-grained emotion distribution of human values. The y-axis represents the value category of the dataset at level 2, and the x-axis shows the fine-grained emotions from the original taxonomy of GoEmotions. The blue color in this heatmap is assigned for emotions under positive sentiment categories, and red and orange indicate emotions of negative and ambiguous categories, respectively, . 6 of 11 Entropy 2024, 26, 327 4. Results and Discussion
id: 22f017b0558649e7161126699ea3d241 - page: 6