DDFI: Diverse and Distribution-aware Missing Feature Imputation via Two-step Reconstruction

2025-12-06

AI Summary

In real-world graph data, node features are often missing due to privacy constraints, severely degrading GNN performance. Existing feature propagation (FP) methods struggle with non-fully-connected graphs, suffer from over-smoothing, and are limited to transductive learning—ignoring feature distribution shift in inductive settings. To address these challenges, we propose DDFI, a two-stage feature imputation framework integrating FP with a graph-structured masked autoencoder (MAE). Its key contributions are: (1) Co-Label Linking, a novel algorithm enabling cross-component feature propagation on weakly connected graphs; (2) a two-step reconstruction mechanism jointly enforcing local consistency and global distribution alignment to mitigate inductive shift and enhance feature diversity; and (3) Sailing, a realistic benchmark dataset featuring naturally missing node features. Extensive experiments across six public benchmarks and Sailing demonstrate that DDFI consistently outperforms state-of-the-art methods under both transductive and inductive settings, significantly improving imputation quality and model robustness.

Technology Category

• Machine Learning: Graph-based Machine Learning
• Search and Optimization: Distributed Search
• Data Mining & Knowledge Management: Graph Mining, Social Network Analysis & Community

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Incomplete node features are ubiquitous in real-world scenarios, e.g., the attributes of web users may be partly private, which causes the performance of Graph Neural Networks (GNNs) to decline significantly. Feature propagation (FP) is a well-known method that performs well for imputation of missing node features on graphs, but it still has the following three issues: 1) it struggles with graphs that are not fully connected, 2) imputed features face the over-smoothing problem, and 3) FP is tailored for transductive tasks, overlooking the feature distribution shift in inductive tasks. To address these challenges, we introduce DDFI, a Diverse and Distribution-aware Missing Feature Imputation method that combines feature propagation with a graph-based Masked AutoEncoder (MAE) in a nontrivial manner. It first designs a simple yet effective algorithm, namely Co-Label Linking (CLL), that randomly connects nodes in the training set with the same label to enhance the performance on graphs with numerous connected components. Then we develop a novel two-step representation generation process at the inference stage. Specifically, instead of directly using FP-imputed features as input during inference, DDFI further reconstructs the features through the whole MAE to reduce feature distribution shift in the inductive tasks and enhance the diversity of node features. Meanwhile, since existing feature imputation methods for graphs only evaluate by simulating the missing scenes with manually masking the features, we collect a new dataset called Sailing from the records of voyages that contains naturally missing features to help better evaluate the effectiveness. Extensive experiments conducted on six public datasets and Sailing show that DDFI outperforms the state-of-the-art methods under both transductive and inductive settings.

Problem

• Imputes missing node features in graphs
• Addresses over-smoothing and distribution shift issues
• Enhances performance on graphs with disconnected components

Innovation

• Combines feature propagation with graph-based Masked AutoEncoder
• Uses Co-Label Linking to connect nodes with same label
• Reconstructs features through MAE to reduce distribution shift