Graph data is often used to model relationships as data. Graph analysis techniques need to be developed from multiple perspectives. Key challenges include prediction of labels and links within a graph, ranking node importance, detection of patterns, and data partitioning. To address these challenges, we aim to develop deep learning techniques, node importance analysis methods, clustering techniques, and patten detection methods for graph data. These advancements will enable more effective graph data analysis.

We do not focus on the utility of graph analysis but also fairness. Fair graph data analysis is expected to develop in various fields, including sociological analysis of human relationships or the relationships between people and professions, as well as industrial applications such as product recommendations in e-commerce. We deepen fair graph analysis.

To evaluate the performance of graph data analysis, we will construct real-world graphs such as knowledge graphs, SNS data, citation networks, and academic collaboration data. Additionally, we will benchmark existing technologies through automation and graph generation techniques. By implementing a framework that facilitates graph deep learning and data mining, we aim to democratize both graph analysis and fair graph analysis.

We will evaluate and clarify the social biases present in real-world graph data from a data science perspective. Furthermore, we will assess the fairness of AI technologies through the lens of philosophy and ethics, examining their societal impact and philosophical implications. Additionally, we will explore the necessity of new evaluation metrics for fairness in graph analysis. Through this approach, we aim to pursue fairness in real-world graphs and AI technologies from both data science and philosophical-ethical perspectives.