A research team led by Professor Kim Sang-wook of the Department of Computer Science at Hanyang University has been awarded the “Best Research Paper Award Honorable Mention” at the 31st ACM KDD 2025, the most prestigious international conference in the field of data science. The award recognizes their development of SIGEM, a novel artificial intelligence (AI)-based network embedding technology.

Networks (graphs) are data structures used to represent complex relationships between objects. Network embedding refers to the process of converting these networks into low-dimensional vectors that are easier for AI models to process. The technology is widely used by global companies such as Meta (Facebook), Amazon, Netflix, and Google in recommendation, search, and personalization services.

However, existing embedding techniques have limitations. They often focus only on local neighbor relationships, failing to capture the overall network structure. In addition, they struggle to learn from sparsely connected nodes, cannot distinguish directional connections, and are often limited to specific types of networks, restricting both accuracy and generalizability.

To address these challenges, Professor Kim’s team developed SIGEM. This new technique precisely calculates the similarity between all pairs of nodes in a network, faithfully reflecting their relationships in the resulting vector representations. It also incorporates edge directionality into the learning process, preserving the global structure of the network. As a result, SIGEM is highly versatile, applicable to both directed and undirected networks.

The team also proposed a novel similarity calculation method, LINOW, which enables fast and accurate similarity estimation even in large-scale networks, enhancing the practical applicability of their approach to real-world complex network analysis.

In benchmark tests conducted on eight real-world networks, SIGEM achieved a 102% improvement in link prediction performance and up to a 21% increase in node classification accuracy, outperforming state-of-the-art techniques. Its consistent performance in large and complex network environments also highlighted its potential for industrial applications.

Professor Kim remarked, “We are honored to receive one of the top three awards out of more than 2,000 submissions at ACM KDD, the leading global conference in data science. This recognition affirms the originality and technological strength of our team.” He added, “SIGEM offers a practical solution for analyzing massive networks in real-world industrial settings.”

This sprit 토토사이트 was supported by the SW StarLab Program and the AI Graduate School Support Program of the Ministry of Science and ICT and the Institute of Information & Communications Technology Planning & Evaluation (IITP). The paper, titled “SIGEM: A Simple yet Effective Similarity-based Graph Embedding Method”, lists sprit 토토사이트 Professor Reyhani Hamedani Masoud as the first author, with sprit 토토사이트ers Oh Jeong-seok and Cho Sung-woon as the second and third authors, respectively. Professor Kim Sang-wook served as the corresponding author.