Span Extraction Method Based on Graph Attention Mechanism for Extraction of Sentiment Triples

Aspect-Based Sentiment Triplet Extraction (ASTE) is an important task in sentiment analysis, which aims to extract triplets containing aspects, opinions, and sentiments from text. This is of great significance for processing sentiment information in large amounts of text data such as consumer reviews and social media content. Existing studies mainly focus on span representation and entity interaction, but often ignore the relationship between spans and the interactive information between entities. To address this problem, this study proposes a sentiment span labeling method based on a graph attention mechanism, which aims to better capture the relationship between spans and effectively use the interactive information between entities for sentiment classification. We also note that the current evaluation of the ASTE task mostly relies on small-scale datasets, which limits the accurate evaluation of the method effect. To this end, we conducted comprehensive experiments on a large dataset. The experimental results show that our method exhibits significant effectiveness and efficiency in a wide range, providing strong support for further research on the sentiment triplet extraction task. In summary, this study not only proposes a novel sentiment span labeling method, but also makes significant progress in experimental evaluation, providing an important reference for the research and application of sentiment triplet extraction tasks.