MMCTR Dataset

该数据集包含了用户与物品互动的数据，同时丰富了多模态特征，如文本和视觉信息。此外，数据集还包括了经过主成分分析（PCA）降维的BERT和CLIP嵌入，这些嵌入在多模态推荐系统中的有效性已经得到了评估。该数据集的任务是进行多模态点击通过率（CTR）的预测。This dataset contains user-item interaction data, and is enriched with multimodal features including text and visual information. Additionally, the dataset provides BERT and CLIP embeddings dimensionality-reduced via Principal Component Analysis (PCA), whose effectiveness in multimodal recommendation systems has been evaluated. The task of this dataset is multimodal Click-Through Rate (CTR) prediction.