A Benchmarking Framework for Network Classification Methods

2025-06-06

AI Summary

This work addresses the challenge of evaluating robustness and generalization of network classification methods under structural noise. We introduce the first synthetic network benchmark framework supporting hierarchical categories (classes/subclasses) with controllable noise types and intensities. To enhance subcategory discrimination under noise, we propose DTWB—a novel feature extraction method integrating deterministic tourist walks (DTW) with dynamic edge-weight adjustment. We systematically compare DTWB against LLNA, DTW, Graph2Vec, and classical topological metrics. Results show DTWB achieves state-of-the-art accuracy on both class- and subclass-level classification tasks and exhibits superior noise robustness; LLNA and DTW follow closely, Graph2Vec performs moderately, while traditional topological features yield the weakest performance. This work establishes a reproducible, hierarchical benchmark for rigorous evaluation of network classification algorithms and provides an effective, noise-resilient feature learning tool.

Technology Category

• Data Mining & Knowledge Management: Graph Mining, Social Network Analysis & Community
• Machine Learning: Graph-based Machine Learning
• Natural Language Processing: Information Extraction

Application Category

• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods
• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics

Abstract

Network classification plays a crucial role in the study of complex systems, impacting fields like biology, sociology, and computer science. In this research, we present an innovative benchmark dataset made up of synthetic networks that are categorized into various classes and subclasses. This dataset is specifically crafted to test the effectiveness and resilience of different network classification methods. To put these methods to the test, we also introduce various types and levels of structural noise. We evaluate five feature extraction techniques: traditional structural measures, Life-Like Network Automata (LLNA), Graph2Vec, Deterministic Tourist Walk (DTW), and its improved version, the Deterministic Tourist Walk with Bifurcation (DTWB). Our experimental results reveal that DTWB surpasses the other methods in classifying both classes and subclasses, even when faced with significant noise. LLNA and DTW also perform well, while Graph2Vec lands somewhere in the middle in terms of accuracy. Interestingly, topological measures, despite their simplicity and common usage, consistently show the weakest classification performance. These findings underscore the necessity of robust feature extraction techniques for effective network classification, particularly in noisy conditions.

Problem

• Evaluating effectiveness of network classification methods
• Testing resilience against structural noise in networks
• Comparing performance of feature extraction techniques

Innovation

• Synthetic benchmark dataset for network classification
• Five feature extraction techniques evaluated
• DTWB outperforms others in noisy conditions