CARE: Causality Reasoning for Empathetic Responses by Conditional Graph Generation

TL;DR

CARE introduces a conditional variational graph auto-encoder (CVGAE) to jointly reason all plausible causalities in empathetic conversations, conditioned on user emotion, dialogue history, and predicted future content. The framework generates a causal graph and infuses its edges into the decoder via a multi-source attention mechanism, enabling deeper understanding of user feelings and experiences. Empirical results on EmpatheticDialogues show state-of-the-art performance in automatic metrics and human judgments, validating the benefits of interdependent causality reasoning for empathetic response generation. The work highlights the importance of considering both experiences and their interrelations to produce more empathetic and relevant responses, with practical implications for socially aware conversational agents.

Abstract

Recent approaches to empathetic response generation incorporate emotion causalities to enhance comprehension of both the user's feelings and experiences. However, these approaches suffer from two critical issues. First, they only consider causalities between the user's emotion and the user's experiences, and ignore those between the user's experiences. Second, they neglect interdependence among causalities and reason them independently. To solve the above problems, we expect to reason all plausible causalities interdependently and simultaneously, given the user's emotion, dialogue history, and future dialogue content. Then, we infuse these causalities into response generation for empathetic responses. Specifically, we design a new model, i.e., the Conditional Variational Graph Auto-Encoder (CVGAE), for the causality reasoning, and adopt a multi-source attention mechanism in the decoder for the causality infusion. We name the whole framework as CARE, abbreviated for CAusality Reasoning for Empathetic conversation. Experimental results indicate that our method achieves state-of-the-art performance.

Figures (4)

• Figure 1: Causality reasoning results of GEE$_{\text{MIME}}$Kim:2021:empathy and our proposed method in a real case. Arrows indicate relations from cause to effect, while strikeout arrows indicate no causal relations. GEE$_{\text{MIME}}$ detects only direct causes and effects of the user's emotion independently, while ours extends the causality scope and reasons causalities interdependently.
• Figure 2: The overview of our proposed framework. The solid lines represent modules or data used for both posterior and prior computation, while the dot lines represent modules or data used only for posterior computation.
• Figure 3: Model performs (BLEU-4) when we gradually increase the number of selected relationships $k$. The solid line and dot line represent BLEU-4 and two period moving average, respectively. For each $k$, we repeat five runs and compute the average BLEU-4.
• Figure 4: This is the user interface of the system for human ratings.