Abstract:The recent integration of Graph Neural Networks (GNNs) into recommendation has led to a novel family of Collaborative Filtering (CF) approaches, namely Graph Collaborative Filtering (GCF). Following the same GNNs wave, recommender systems exploiting Knowledge Graphs (KGs) have also been successfully empowered by the GCF rationale to combine the representational power of GNNs with the semantics conveyed by KGs, giving rise to Knowledge-aware Graph Collaborative Filtering (KGCF), which use KGs to mine hidden user intent. Nevertheless, empirical evidence suggests that computing and combining user-level intent might not always be necessary, as simpler approaches can yield comparable or superior results while keeping explicit semantic features. Under this perspective, user historical preferences become essential to refine the KG and retain the most discriminating features, thus leading to concise item representation. Driven by the assumptions above, we propose KGUF, a KGCF model that learns latent representations of semantic features in the KG to better define the item profile. By leveraging user profiles through decision trees, KGUF effectively retains only those features relevant to users. Results on three datasets justify KGUF's rationale, as our approach is able to reach performance comparable or superior to SOTA methods while maintaining a simpler formalization. Link to the repository:

What problem does this paper attempt to address?

### Problems the Paper Attempts to Solve This paper aims to address some key issues in recommendation systems and proposes a new knowledge-aware graph recommendation algorithm called KGUF (Knowledge Graph User-based Filtering). #### Main Issues 1. **Complexity of User Intent Modeling**: Existing knowledge-aware graph collaborative filtering (KGCF) methods often require complex aggregation strategies to learn meaningful user and item features, making these methods difficult to implement in practical applications. 2. **Nuances in User Preferences**: Traditional collaborative filtering (CF) and graph collaborative filtering (GCF) methods tend to overlook the details of user preferences related to specific item characteristics, leading to insufficient personalized recommendation capabilities. 3. **Simplifying Models to Maintain Performance**: The paper argues that existing architectures can be greatly simplified while maintaining or improving performance, particularly by utilizing content information from knowledge graphs to describe item representations rather than directly using it to construct user features. #### Solutions 1. **KGUF Model**: KGUF is a knowledge graph-based recommendation algorithm that learns embeddings of users and items through linear propagation on a collaborative graph and directly integrates the latent representations of semantic features into item representations. 2. **Decision Tree Mechanism**: KGUF uses a decision tree mechanism to select the most relevant semantic features from the knowledge graph and filter out noise features that contribute little to item descriptions but significantly impact computational costs. 3. **Knowledge-Aware Aggregation Scheme**: KGUF proposes a simple yet effective knowledge-aware aggregation scheme that combines collaborative signals and knowledge signals, thereby improving the representations of items and users. ### Experimental Results and Discussion 1. **Overall Performance**: KGUF outperforms or is comparable to the current best methods on three datasets, particularly excelling on the MovieLens 1M dataset. 2. **Knowledge Contribution Weight**: By adjusting the α parameter, the optimal balance between content and collaborative contributions (40% content, 60% collaborative) is found, and removing knowledge information alone has a greater impact on model performance than removing collaborative information. 3. **Decision Tree Construction**: Negative sampling enhancement and the maximum depth of the tree have a significant impact on overall recommendation quality. Increasing the number of negative samples can improve accuracy, but excessive increases do not lead to continuous improvement. The maximum depth of the tree also needs to be adjusted according to the specific dataset.

KGUF: Simple Knowledge-aware Graph-based Recommender with User-based Semantic Features Filtering

KGCFRec: Improving Collaborative Filtering Recommendation with Knowledge Graph

Knowledge-Enhanced Recommendation with User-Centric Subgraph Network

Knowledge Graph Enhanced Neural Collaborative Filtering with Residual Recurrent Network

Knowledge graph fine-grained network with attribute transfer for recommendation

Sparse Feature Factorization for Recommender Systems with Knowledge Graphs

Knowledge Graph Convolutional Networks for Recommender Systems

RGCF: Refined Graph Convolution Collaborative Filtering with concise and expressive embedding

KGIE: Knowledge graph convolutional network for recommender system with interactive embedding

KGCNA: Knowledge Graph Collaborative Neighbor Awareness Network for Recommendation

Attentive Knowledge-aware Graph Convolutional Networks with Collaborative Guidance for Personalized Recommendation

MetaKRec: Collaborative Meta-Knowledge Enhanced Recommender System

Knowledge-aware fine-grained attention networks with refined knowledge graph embedding for personalized recommendation

KQGC: Knowledge Graph Embedding with Smoothing Effects of Graph Convolutions for Recommendation

Knowledge Graph Enhanced Neural Collaborative Recommendation.

DiffKG: Knowledge Graph Diffusion Model for Recommendation

Feature interactive graph neural network for KG-based recommendation

Multi-contrastive Learning Recommendation Combined with Knowledge Graph

Kgat: Knowledge Graph Attention Network For Recommendation

Unifying Knowledge Graph Learning and Recommendation: Towards a Better Understanding of User Preferences

An adaptive category-aware recommender based on dual knowledge graphs