Enhancing Sentiment Analysis of Art Works through Big Data and Multi-Modality Integration

Traditionally, AI classification has relied on single-modality networks, which could overlook valuable complementary information from other modalities, leading to potential inaccuracies or incomplete interpretations. This paper introduces the concept of big data and its transformative role in bolstering the accuracy and comprehensiveness of sentiment analysis. By harnessing the power of large-scale datasets, the model can delve into the intricate details of complex emotional nuances in art. This innovative approach contributes to a more profound understanding of art by capturing the myriad of emotions embedded in artistic expressions. The expansive and diverse nature of big data enables the model to capture a wider spectrum of emotions, leading to more nuanced and accurate interpretations of art. We propose an innovative emotional classification model that harnesses big data integration and multimodal fusion techniques, combining image and text modalities to enhance comprehension, facilitate dissemination, and foster deeper art appreciation. Utilizing the Wolfflin's Affective Sentiment Analysis (WASA) dataset, the model effectively classifies complex emotional nuances in artworks. Despite the visual modality's limited accuracy due to subjective interpretation and small dataset size, the text modality achieves higher accuracy, indicating its strength in expressing emotional features. The fusion technology, powered by big data, explores the importance of each modality, significantly improving classification accuracy. This innovative dual-modality fusion, aided by the extensive scope and depth of big data, enables a more comprehensive understanding of the emotional content in artworks. This model eliminates the need for domain-specific expertise, offering a versatile solution that can adapt across different artistic domains, styles, and genres. The dataset used in this research is publicly available at: https://www.jianguoyun.com/p/DS7GzoQQkOuIDBjOi6IFIA.