Visual Analytics of Neighborhood Attribute Profiles for Exploring Structural Equivalence
Authors
Kohei Arimoto (Teikoku Databank, Ltd.), Masahiko Itoh (Hokkaido Information University)
Presentation
- Session
- Connecting the Dots
- Time
- Wednesday, Nov 11, 13:09 – 13:18 (US/Eastern) · session 13:00 – 14:30
- Location
- Hall Essex north
Links
Sign in to access the preprint PDF.
Sign in- Download Supplemental Material
Keywords
Visual Analytics, Attributed Networks, Structural Equivalence, Dimensionality Reduction.
Abstract
Exploring similar nodes in attributed networks represents a key challenge in data mining. While recent representation learning methods embed networks into low-dimensional vectors, they often implicitly assume a uniform and continuous feature space. This paper proposes a visual analytics approach using dimensionality reduction to help clarify the true topological structure of high-dimensional feature spaces formed by nodes' neighborhood attribute profiles. Analyzing inter-firm transaction networks indicates that structural roles can form complex, non-linear manifolds with density biases. Comparing this feature space with industry classifications suggested: (1) supply chain hierarchies transition continuously; (2) categories treated identically under general semantics can be clearly separated by actual transaction networks; and (3) a single industry label may fragment into multiple regions. These findings suggest potential limitations in assuming identical semantics imply similar structural roles and highlight the possible need for new similarity metrics aligned with manifold topology.
For Practitioners
Data scientists, machine learning engineers (especially those working with graph representation learning), financial/credit analysts, business intelligence (BI) practitioners, and supply chain analysts. Practitioners can apply the insights and methods from this paper in two primary ways: Improving Graph Machine Learning Models: Data scientists and ML engineers can apply the proposed neighborhood attribute profiling and L1 normalization techniques to mitigate the bias caused by hub nodes in scale-free networks. Furthermore, they can use this UMAP-based visual analytics approach to intuitively debug and validate the geometric validity of latent feature spaces, rather than implicitly trusting uniform linear assumptions or static attribute labels. Discovering True Business Relationships: Financial analysts and BI practitioners can apply these visual techniques to real-world corporate data to prospect new B2B customers or identify true structural competitors. By understanding that actual transaction topologies often diverge from conventional semantic classifications (e.g., finding fragmented roles within a single industry label), practitioners can perform more accurate market segmentations and risk assessments that reflect the reality of complex supply chains.