Abstract:Large Language Models (LLMs) have recently displayed their extraordinary capabilities in language understanding. However, how to comprehensively assess the sentiment capabilities of LLMs continues to be a challenge. This paper investigates the ability of LLMs to detect and react to sentiment in text modal. As the integration of LLMs into diverse applications is on the rise, it becomes highly critical to comprehend their sensitivity to emotional tone, as it can influence the user experience and the efficacy of sentiment-driven tasks. We conduct a series of experiments to evaluate the performance of several prominent LLMs in identifying and responding appropriately to sentiments like positive, negative, and neutral emotions. The models' outputs are analyzed across various sentiment benchmarks, and their responses are compared with human evaluations. Our discoveries indicate that although LLMs show a basic sensitivity to sentiment, there are substantial variations in their accuracy and consistency, emphasizing the requirement for further enhancements in their training processes to better capture subtle emotional cues. Take an example in our findings, in some cases, the models might wrongly classify a strongly positive sentiment as neutral, or fail to recognize sarcasm or irony in the text. Such misclassifications highlight the complexity of sentiment analysis and the areas where the models need to be refined. Another aspect is that different LLMs might perform differently on the same set of data, depending on their architecture and training datasets. This variance calls for a more in-depth study of the factors that contribute to the performance differences and how they can be optimized.

Sentiment Analysis Based on Chinese Thinking Modes

Analysis Based on Chinese Thinking Modes

A Mixture Language Model for the Classification of Chinese Online Reviews.

Reflections on Sentiment/Opinion Analysis

Topic-related Chinese Message Sentiment Analysis

Improving Chinese Sentiment Analysis via Segmentation-Based Representation Using Parallel CNN.

Using Bilingual Lexicon to Judge Sentiment Orientation of Chinese Words

Sentiment interpretability analysis on Chinese texts employing multi-task and knowledge base

Implicit sentiment analysis based on multi-feature neural network model

Sentiment Analysis of Text Based on Three-Way Decisions.

Sentiment Analysis of Chinese Words Using Word Embedding and Sentiment Morpheme Matching.

A Syntax-based BSGCN Model for Chinese Implicit Sentiment Analysis with Multi-classification

Sentiment Classification Using Negative and Intensive Sentiment Supplement Information

Sentiment Classification for Chinese News Using Machine Learning Methods

Exploring the Emotional Characteristics of Chinese Language Literature Based on the Thinking Operation Model

Dimensional Sentiment Analysis of Traditional Chinese Words Using Pre-Trained Not-Quite-Right Sentiment Word Vectors and Supervised Ensemble Models

Do Large Language Models Possess Sensitive to Sentiment?

Sentiment Classification in Turn-Level Interactive Chinese Texts of E-learning Applications

Chinese Sentential Semantic Mode and Verification

Sentiment Analysis for Short Chinese Text Based on Character-Level Methods.

Sentiment Analysis of Modern Chinese Literature Based on Deep Learning