MUMBAI, India, Aug. 10 -- Intellectual Property India has published a patent application (202641080364 A) filed by Christ University on June 30, 2026, for A Method Of Characterisation Of Higher Order Graphs Into Clusters For Efficient study of Graph th.
Inventors include Rajshree Dahal; Hari Baskar R; and Debabrata Samanta.
The application for the patent was published on July 31, 2026, under issue no. 31/2026.
Abstract: The present invention relates to a novel graph-analytic framework and computational method for characterizing complex datasets using cluster-based graph representations and domination- theoretic structural analysis. The invention provides a systematic technique for mapping high dimensional data to a graph structure, discovering clusters in the graph, building a higher level cluster graph, and using domination-based parameters to extract structural, connectivity, coverage and stability properties of the underlying network. In one embodiment, a dataset is first represented as a similarity graph based on neighbourhood relationships, distance measures or other affinity criteria. Then a clustering mechanism such as k-means, spectral clustering, hierarchical clustering or equivalent method is used to divide the graph into clusters. Each cluster is then represented as a supemode, and inter-cluster relationships are utilized to construct a reduced cluster graph that preserves the essential topological characteristics of the original network. The invention further comprises the application of domination-theoretic parameters such as (but not limited to) domination, connected domination, convex domination, nonsplit domination, nonsplit convex domination, and other related graph invariants to analyze the cluster graph. These parameters provide quantitative measures o f the coverage, influence, robustness, fault-tolerance, redundancy and structural cohesion of the network. The proposed method can significantly reduce the computational complexity while preserving the meaningful structural information by working on the cluster graph instead of the original graph. The provided framework can be used in different areas including machine learning, artificial intelligence, communication networks, sensor systems, transportation networks, social networks, biological networks, cybersecurity, and large-scale data analytics. The experimental validation on benchmark datasets shows that the method is able to identify critical clusters, determine the optimal representative structures and assess the network resilience. Thus, the invention offers a unified computational paradigm that integrates clustering, graph abstraction, and domination-based analysis to efficiently characterize and optimize complex interconnected systems.
Disclaimer: Curated by HT Syndication.
