Evaluating Emotional Nuances In Dialogue Summarization

“Affective content has been the target of a few summarization tasks such as opinion summarization [Wang and Ling, 2016]. However, opinion is only a subset of affective expressions and such task mainly focuses on non-dialogue texts reviews. In dialogues, factual information and subjective content are often intertwined. While it is essential to summarize the most pertinent factual information, the subjective content can also provide valuable insights, for example, improve customer service interactions and better support patients in the context of e-healthcare. Therefore, it is important to extract and synthesize this content when summarizing dialogues, as it can offer numerous benefits. Even though summarizing the affective part of dialogues could be highly valuable in many applications including health care, call centers, meetings or collaborative work, it has been understudied, with only one study focusing on this topic [Roman et al., 2008]. For instance, in collaborative interactions scenarios, participants usually take turns expressing their opinion, whether they agree or disagree with each other’s proposals, and what they think. In the customer service domain, call center telephone conversations are often used for monitoring purpose and to improve service quality. Many calls contain emotional information that are deemed important to report [Roman et al., 2008]. However, it is not unusual that when summarization tasks are designed, summarization guidelines often focus on facts and goals and leave human summarizers without concrete way to deal with emotional content [Roman et al., 2008]. A recent counter-example is the DialogSum dataset [Chen et al., 2021], where annotators were required to pay extra attention to several different aspects including Emotions. This shows that the community acknowledges more and more that it is an important piece of information.”