Abstract: Sentiment quadruple extraction, which aims to identify holders, opinions, aspects, and sentiment polarities, is an important task in aspect-based sentiment analysis(ABSA). Its objective is to extract complete sentiment structures consisting of holders, opinions, aspects, and sentiment polarities(positive, negative, neutral) from review texts, thereby providing a more comprehensive information. Existing ABSA methods often overlook the role of sentiment holders and struggle to handle nested structures within sentiment quadruples. To address these issues, a biaffine dependency parsing-based sentiment quadruple extraction model, termed BDSQE, is proposed. The model first employs a U-Net architecture to enhance contextual word representations. It then predicts sentiment polarity labels through a classification module and utilizes biaffine dependency parsing to identify structural relationships among sentiment elements. Based on the parsed dependencies and polarity labels, holders, opinions, and aspects are jointly extracted to construct sentiment quadruples. By explicitly modeling sentiment holders and dependency relations, BDSQE enriches sentiment representations and effectively alleviates the nested structure problem encountered in existing approaches. Experimental results on five benchmark datasets demonstrate that BDSQE outperforms competing methods on most evaluation metrics.