Recently, multimodal recommendation (MMRec) has received much attention, which models user preferences based on both user behaviors and modality information. Although current graph neural network based methods yield notable results in MMRec, certain limitations persist among these methods. 1) Most methods rely on pre-trained networks to extract modality features but fail to remove modality noise. 2) Recent methods leverage InfoNCE strategy to align representation, while ignoring the effect of feature redundancy and lacking sufficient alignment between different modality features. Such limitations ultimately harm the recommendation performance. To this end, we propose a Dual Mutual Information-Driven Multimodal Recommendation Model with Denoising Graph Autoencoder (DMIGA). Specifically, to reduce the noise within modality features, we design a denoising graph autoencoder with a cross-modal consistency constraint. Furthermore, we propose a dual mutual information learning mechanism on both feature and instance levels, to reduce the feature redundancy and align different representations. Experimental results on three real-world datasets consistently demonstrate that DMIGA outperforms state-of-the-art methods, with an average of 3.8% improvement.

Recommendation is a key mechanism for modern users to access items of their interests from massive entities and information. Recently, graph contrastive learning (GCL) has demonstrated satisfactory results on recommendation, due to its ability to enhance representation by integrating graph neural networks (GNNs) with contrastive learning. However, those methods often generate contrastive views by performing random perturbation on edges or embeddings, which is likely to bring noise in representation learning. Besides, in all these methods, the degree of user preference on items is omitted during the representation learning process, which may cause incomplete user/item modeling. To address these limitations, we propose the Adaptive Fusion of Multi-View Graph Contrastive Recommendation (AMGCR) model. Specifically, to generate the informative and less noisy views for better contrastive learning, we design four view generators to learn the edge weights focusing on weight adjustment, feature transformation, neighbor aggregation, and attention mechanism, respectively. Then, we employ an adaptive multi-view fusion module to combine different views from both the view-shared and the view-specific levels. Moreover, to make the model capable of capturing preference information during the learning process, we further adopt a preference refinement strategy on the fused contrastive view. Experimental results on three real-world datasets demonstrate that AMGCR consistently outperforms the state-of-the-art methods, with average improvements of over 10% in terms of Recall and NDCG. Our code is available on https://github.com/Du-danger/AMGCR.