Recognizing Emotion Cause in Conversations

Abstract

We address the problem of recognizing emotion cause in conversations, define two novel sub-tasks of this problem, and provide a corresponding dialogue-level dataset, along with strong transformer-based baselines. The dataset is available at https:// github. com/ decla re-lab/ RECCON. Recognizing the cause behind emotions in text is a fundamental yet underexplored area of research in NLP. Advances in this area hold the potential to improve interpretability and performance in affect-based models. Identifying emotion causes at the utterance level in conversations is particularly challenging due to the intermingling dynamics among the interlocutors. We introduce the task of Recognizing Emotion Cause in CONversations with an accompanying dataset named RECCON, containing over 1,000 dialogues and 10,000 utterance cause/effect pairs. Furthermore, we define different cause types based on the source of the causes, and establish strong Transformer-based baselines to address two different sub-tasks on this dataset. Our transformer-based baselines, which leverage contextual pretrained embeddings, such as RoBERTa, outperform the state-of-the-art emotion cause extraction approaches on our dataset. We introduce a new task highly relevant for (explainable) emotion-aware artificial intelligence: recognizing emotion cause in conversations, provide a new highly challenging publicly available dialogue-level dataset for this task, and give strong baseline results on this dataset.

Towards an Online Empathetic Chatbot with Emotion Causes

“…It is reasonable because ECE is a more difficult task than ECf. As discussed in [22], emotion causes can be context-depedent or context-independent, and sometimes they are latent and unmentioned in the text. Inspired by previous work [4,33], we add up the losses from the two tasks and joint learning the model parameters.…”

“…Emotion Cause. Following previous work [22,23], we define emotion cause as the continuous text span in an utterance that can be used to detected or inferred the speaker's emotion.…”

“…To facilitate empathetic responding, we argue that it is also necessary for the chatbots to learn the causes behind user's emotion, a.k.a., emotion causes, in addition to the emotion class. Although emotion cause has been regarded as an important information for emotion classification in conversation [12,22,23] and news text [11,33], its benefit for empathetic response generation is yet to be explored. Especially for online chatbots, it is non-trivial to exploit the causal emotion information.…”

TSAM: A Two-Stream Attention Model for Causal Emotion Entailment

“…(Poria et al, 2017(Poria et al, , 2019bWang et al, 2020;. However, Poria et al Figure 1: Example conversations sampled from the benchmark dataset (Poria et al, 2021) (2021) point out that these studies lack further reasoning about emotions, such as understanding the stimuli and the cause of the emotion. Since Recognizing Emotion Cause in Conversations (REC-CON) holds the potential to improve the interpretability and performance of affect-based models, Poria et al (2021) put forward a new promising task, named RECCON, which includes two different sub-tasks: Causal Span Extraction (CSE) at word/phrase level and Causal Emotion Entailment (CEE) at utterance level.…”

EmoWOZ: A Large-Scale Corpus and Labelling Scheme for Emotion Recognition in Task-Oriented Dialogue Systems

“…In task-oriented dialogues, emotion is centred around the user goal, making it more contextual and subtle. Therefore, besides inferring emotional states from dialogue utterances, an agent also needs to reason about emotion-generating situations (Poria et al, 2021).…”