<style>
img {
display: block;
margin-left: auto;
margin-right: auto;
}
</style>

> [Paper link](https://arxiv.org/pdf/2009.09708.pdf) | [Note link](https://zhuanlan.zhihu.com/p/494015775) | [Code link](https://github.com/qtli/KEMP) | AAAI 2022

:::success
**Thoughts**
:::

## Abstract

They propose to leverage external knowledge, including commonsense knowledge and emotional lexical knowledge, to explicitly understand and express emotions in empathetic dialogue generation.
- Enrich the dialogue history by jointly interacting with external knowledge and construct an emotional context graph
- Learn emotional context representations from the knowledge-enriched emotional context graph and distill emotional signals
- Propose an emotional cross-attention mechanism to learn the emotional dependencies from the emotional context graph

## Introduction

Studies on social psychology suggest that empathy is a crucial factor towards a more humanized dialogue system.

Humans usually rely on experience and external knowledge to acknowledge and express implicit emotions. To exploit this phenomenon more concretely, they quantitatively investigate effects of external knowledge in understanding emotions on an empathetic dialogue corpus, EMPATHETICDIALOGUES.


During the investigations, they observe another phenomenon that emotional dependency and emotional inertia commonly appear with external knowledge in empathetic conversations.

## Related Work

**Emotional dialogue generation**

**Empathetic dialogue generation**

## Preliminaries

## Method

### Emotional context graph

### Emotional context encoder

### Emotion-dependency decoder

## Experimental Settings

## Results and analysis

## Conclusion and outlook