A Knowledge Enhanced Pre-Training Model for Chinese Weibo Sentiment Analysis
Li Yao · 2024
Sentiment analysis aims to automatically identify and extract subjective information such as tendencies, stances, evaluations, and opinions from text. Current state-of-the-art Natural Language Processing (NLP) methods, primarily based on large-scale pre-trained language models (PLM), have significantly advanced the field of sentiment analysis. However, these methods often struggle due to the scarcity of annotated data. To address these challenges, we present a knowledge-enhanced pre-trained model for the Chinese social media platform. This model employs a phrase-based masking strategy and incorporates knowledge-related information by making adjustments to the structure of the intermediate encoder layer. We validated the model's performance through comparative experiments with other baseline methods, demonstrating the effectiveness of integrating sentiment resources into pre-trained language models.