Visual analytics of genealogy with attribute-enhanced topological clustering

Clustering is able to present a brief illustration for families of interest and patterns of significance within large-scale genealogical datasets. In the traditional clustering methods, topological features are mostly taken for summarizing and organizing family trees. However, plentiful attributes are ignored which are also important to enhance the understanding and interpretation of genealogical clustering features. Thus, it is a crucial task to combine structures and attributes into a clustering model for exploring genealogy datasets. In this paper, we propose an attribute-enhanced topological clustering method for exploring genealogy datasets based on partial least squares (PLS). Firstly, a graphlet kernel method is utilized to measure the structure difference between family trees. Then, we leverage PLS to combine the learned vectors and multiple attributes, and a joint dimensionality reduction method is applied to project the high-dimensional vectors into a two-dimensional space in which a distance-based clustering method is employed to aggregate the similar family trees taking both the topological structures and attribute features into consideration. Further, we implement a visual analysis system with multi-view collaboration, including glyph, family tree view and parallel coordinate view, to represent, evaluate and explore the clustering features. Case studies and quantitative comparisons based on real-world genealogy datasets have demonstrated the effectiveness of our method in genealogical clustering and exploration. Graphic abstract