GoEmotions: A Dataset of Fine-Grained Emotions

Demszky, Dorottya; Movshovitz-Attias, Dana; Ko, Jeongwoo; Cowen, Alan; Nemade, Gaurav; Ravi, Sujith

doi:10.48550/arxiv.2005.00547

Cited by 50 publications

(116 citation statements)

References 11 publications

Supporting

Mentioning

115

Contrasting

Order By: Relevance

“…The classes of lowest prevalence, such as scared, had the poorest results, while the more frequent classes, such as adoring, approving, and saddened had the highest results. To put these results in further perspective, we note that they are on par with applying BERT-based models to the related task of emotion detection in Demszky et al (2020). Specifically, using similar hyper-parameters, that work achieved a macro-averaged F1-score of 0.64 for a taxonomy of 6 labels.…”

Section: Creating and Evaluating Care-bertmentioning

confidence: 84%

“…CARE-BERT provides strong baseline performance for the task of predicting affective response, on par with the SOTA models for emotion recognition. Furthermore, we show that CARE-BERT can be used for transfer learning to a different emotion-recognition task, achieving similar performance to Demszky et al (2020) which relied on manually-labeled training data.…”

Section: Introductionmentioning

confidence: 85%

“…Additionally, the scope of emotions predicted by these works is usually fairly small, ranging from two to five, and most datasets utilized are usually smaller than 10k (making it unclear if they extrapolate well). Among the ML approaches, many SOTA works employ deep learning methods such as LSTM, RNN, DNN (Demszky et al, 2020;Felbo et al, 2017;Barbieri et al, 2018;Huang et al, 2019a;Baziotis et al, 2017;Huang et al, 2019b). However, while many of the deep learning models have shown significant improvement over prior techniques, the results are highly uninterpretable and typically require prohibitively large human-labeled datasets.…”

Section: Emotion Detection In Textmentioning

confidence: 99%

“…In this section we apply our method to social media posts and validate these annotations using human evaluation (Section 4.1). Section 4.2 discusses class-wise error analysis, and in Section 4.3, we explore the alternative possibility of creating CARE db using a SOTA publisher-affect classifier (Demszky et al, 2020) to label the comments.…”

Section: Evaluating Care Dbmentioning

confidence: 99%

“…Here we show that using CARE patterns and the CARE lexicon performs well compared to using SOTA emotion classifiers. Let us define CARE G , a modified version of CARE where steps 1 and 2 are replaced with labels using the GoEmotions classifer, a BERT-based model trained using the GoEmotions dataset (Demszky et al, 2020). After applying the GoEmotions classifier to all comments of the posts with human annotations in CARE db (Section 4.1), we convert the predicted GoEmotions labels using the mapping between the two taxonomies shown in Table 3.…”

Section: Can We Leverage Emotion Classification?mentioning

confidence: 99%

See 4 more Smart Citations

The CARE Dataset for Affective Response Detection

Jane¹,

Halevy²

2022

Preprint

View full text Add to dashboard Cite

Social media plays an increasing role in our communication with friends and family, and our consumption of information and entertainment. Hence, to design effective ranking functions for posts on social media, it would be useful to predict the affective response to a post (e.g., whether the user is likely to be humored, inspired, angered, informed). Similar to work on emotion recognition (which focuses on the affect of the publisher of the post), the traditional approach to recognizing affective response would involve an expensive investment in human annotation of training data.We introduce CARE db , a dataset of 230k social media posts annotated according to 7 affective responses using the Common Affective Response Expression (CARE) method. The CARE method is a means of leveraging the signal that is present in comments that are posted in response to a post, providing high-precision evidence about the affective response of the readers to the post without human annotation. Unlike human annotation, the annotation process we describe here can be iterated upon to expand the coverage of the method, particularly for new affective responses. We present experiments that demonstrate that the CARE annotations compare favorably with crowdsourced annotations. Finally, we use CARE db to train competitive BERT-based models for predicting affective response as well as emotion detection, demonstrating the utility of the dataset for related tasks.

show abstract

Section: Creating and Evaluating Care-bertmentioning

confidence: 84%

Section: Introductionmentioning

confidence: 85%

Section: Emotion Detection In Textmentioning

confidence: 99%

Section: Evaluating Care Dbmentioning

confidence: 99%

Section: Can We Leverage Emotion Classification?mentioning

confidence: 99%

See 3 more Smart Citations

The CARE Dataset for Affective Response Detection

Jane¹,

Halevy²

2022

Preprint

View full text Add to dashboard Cite

show abstract

Data Analysis Using NLP to Sense Human Emotions Through Chatbot

Rahat

Mim

Mahmud

et al. 2022

Communications in Computer and Information Science

View full text Add to dashboard Cite

Comparing User Perspectives in a Virtual Reality Cultural Heritage Environment

Bulla

Giorgis

Gangemi

et al. 2023

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Virtual reality enables the creation of personalized users experiences that put together people of different cultures and ethnicity. We consider a novel concept of virtual reality innovation in museums, which is cognitively grounded, and based on data and their semantics, to enable users to share their experiences, as well as to assume the perspective of other users, with the ultimate goal of increasing social cohesion. The implementation of this scenario requires an autonomous artificial system capable to detect emotions and values from a dialogue involving museum visitors who express their personal point of view, listen to the point of view of other visitors, and assume others' perspectives. An important feature of this system is the ability of detecting similarity and dissimilarity between users' perspectives expressed by speech when exposed to artworks. This aids in defining an effective strategy for sharing diverse user perspectives for increasing social cohesion. Moreover, it enables an unbiased quantification of the success of the interaction in terms of change in the user perspective. Based on results from previous work, we employ the Ekman emotion model and Haidt moral value model to detect emotional and moral value profiles from user descriptions of artworks. We propose a novel method for measuring the similarity between user perspectives by comparing emotional and moral value profiles. Our results show that the employment of unsupervised text classification models is a promising research direction for this task.

show abstract

GoEmotions: A Dataset of Fine-Grained Emotions

Cited by 50 publications

References 11 publications

The CARE Dataset for Affective Response Detection

The CARE Dataset for Affective Response Detection

Data Analysis Using NLP to Sense Human Emotions Through Chatbot

Comparing User Perspectives in a Virtual Reality Cultural Heritage Environment

Contact Info

Product

Resources

About