The field of natural language processing is moving towards more fine-grained sentiment analysis, with a focus on aspect-based sentiment analysis. This involves identifying specific themes or aspects within text that are associated with particular opinions, allowing for deeper insights than traditional sentiment analysis. Recent work has explored the use of large language models for aspect-based sentiment analysis, including the development of new datasets and techniques for quantifying uncertainty. These advancements have the potential to improve the accuracy and robustness of sentiment analysis systems, particularly in domains where labeled data is scarce. Noteworthy papers include:

• A study on the ability of language models to understand suspense in stories, which found that while models can identify certain facets of suspense, they do not process it in the same way as human readers.
• The introduction of EduRABSA, a public dataset for aspect-based sentiment analysis in the education domain, which addresses the need for high-quality datasets in this area.
• A proposal for combining multiple chain-of-thought agents with uncertainty quantification for aspect-category sentiment analysis, which demonstrates the potential for large language models to improve sentiment analysis in label-scarce conditions.
• The creation of a retail-corpus for aspect-based sentiment analysis with large language models, which establishes a baseline for the performance of these models in this domain.