Yeonsung JUNG of the Department of Electronic, Electrical & Computer Engineering, and Woojoong LEE of the Department of Mathematics, received awards for superior presentations from the Korean Institute of Information Scientists and Engineers.
Yeonsung JUNG presented his thesis, "Performance Analysis of Node Embedding Algorithms on Web Graphs". Node embedding has received attention as a very effective representation method while conducting supervised learning on graphic data. This method shows how to define and select the neighbor node of each nodes, and express the connectivity and structural role of each node. Since recent algorithms target the nodes according to their order of priority, it is assumed that performance would vary according to the type of data. To test this thesis, it was compared and analyzed to determine if there is superior performance with node embedding algorithms in terms of node classification using web graphic data from real search engines.
Woojoong LEE presented his thesis entitled "Graph Clustering with Link Structure and Node Attributes". He and Professor Joyce Jiyoung Whang, developed a new graph clustering algorithm which utilized both node features and link structure. By recognizing the similarity between nodes and other nodes, a new graph is formed by adding an edge between nodes with a high degree of similarity. When the newly generated graph was applied to the existing community detection algorithm, better graph clustering was conducted.