Abstract:This paper shows how LLMs (Large Language Models) may be used to estimate a summary of the emotional state associated with piece of text. The summary of emotional state is a dictionary of words used to describe emotion together with the probability of the word appearing after a prompt comprising the original text and an emotion eliciting tail. Through emotion analysis of Amazon product reviews we demonstrate emotion descriptors can be mapped into a PCA type space. It was hoped that text descriptions of actions to improve a current text described state could also be elicited through a tail prompt. Experiment seemed to indicate that this is not straightforward to make work. This failure put our hoped for selection of action via choosing the best predict ed outcome via comparing emotional responses out of reach for the moment.

What problem does this paper attempt to address?

The problem this paper attempts to address is how to use large language models (LLMs) to estimate the emotional state of text and explore the application of these models in constructing synthetic consciousness. Specifically, the authors aim to achieve this goal through the following methods: 1. **Estimation of Emotional Probability Vectors**: By constructing an emotional lexicon and calculating the probability of these emotional words appearing after a given text, an emotional probability vector is generated. This allows for a quantitative assessment of the emotional content of the text. 2. **Application of Sentiment Analysis**: The effectiveness of the proposed method is validated through sentiment analysis of Amazon product reviews. The authors selected 271 words describing emotions to estimate the probability of these words appearing in specific texts. 3. **Preliminary Exploration of Synthetic Consciousness**: Attempts are made to guide the behavior selection of synthetic consciousness through emotional probability vectors. The authors hope to use emotional probability vectors to judge the pros and cons of different behaviors or states, thereby providing decision-making basis for synthetic consciousness. However, the experimental results show that while it is possible to effectively estimate the emotional state of text through emotional probability vectors, generating specific action suggestions from LLMs using tail prompts is not easily achievable. This limits the authors' initial idea of selecting the best behavior targets through emotional probability vectors. Overall, the main contribution of this paper is to propose a new method for estimating the emotional state of text using LLMs and to explore its potential application in the field of synthetic consciousness.

Towards Emotion-Based Synthetic Consciousness: Using LLMs to Estimate Emotion Probability Vectors

Emotionally Numb or Empathetic? Evaluating How LLMs Feel Using EmotionBench

Recognizing Emotion Regulation Strategies from Human Behavior with Large Language Models

Large Language Models Understand and Can be Enhanced by Emotional Stimuli

Contextual Emotion Estimation from Image Captions

Language Models as Emotional Classifiers for Textual Conversation

Lightly-supervised Utterance-Level Emotion Identification Using Latent Topic Modeling of Multimodal Words.

Emotion-Aware Response Generation Using Affect-Enriched Embeddings with LLMs

Emotional intelligence of Large Language Models

Recent Advancement of Emotion Cognition in Large Language Models

Speak From Heart: An Emotion-Guided LLM-Based Multimodal Method for Emotional Dialogue Generation

Modeling Emotions and Ethics with Large Language Models

EVOLVE: Emotion and Visual Output Learning via LLM Evaluation

Can Generative Agents Predict Emotion?

From Text to Emotion: Unveiling the Emotion Annotation Capabilities of LLMs

Revise, Reason, and Recognize: LLM-Based Emotion Recognition via Emotion-Specific Prompts and ASR Error Correction

EmotionPrompt: Leveraging Psychology for Large Language Models Enhancement via Emotional Stimulus

Retrieving Implicit and Explicit Emotional Events Using Large Language Models

Conditioning LLMs with Emotion in Neural Machine Translation

DeepPavlov at SemEval-2024 Task 3: Multimodal Large Language Models in Emotion Reasoning

VLLMs Provide Better Context for Emotion Understanding Through Common Sense Reasoning