MUMBAI, India, Sept. 28 -- Intellectual Property India has published a patent application (202621079911 A) filed by Parul University Parul Institute Of Engineering Technology on June 29, 2026, for System And Method For Automated Academic Influence Archetype Classification Using Graph Convolutional Networks And Shap-Based Interpretability.
Inventor includes Dr. Sanjay Agal.
The application for the patent was published on September 25, 2026, under issue no. 39/2026.
Abstract: A computer-implemented system identifies and classifies academic influence archetypes in educational social networks. A feature extraction module computes a composite influence index from structural features (betweenness centrality, eigenvector centrality, multilayer participation coefficient), behavioral features (temporal centrality trends, network position volatility), and outcome features (publication tier, supervisory success, grant acquisition). A semi-supervised Graph Convolutional Network classifies actors using partial ground truth labels (at most 35% of population). An unsupervised clustering module employs UMAP dimensionality reduction and hierarchical agglomerative clustering to identify five archetypes: Integrative Thought Leader, Specialized Authority, Boundary Spanner, Mentorship Catalyst, and Latent Influencer. A SHAP-based interpretability module identifies multilayer participation coefficient (0.24), betweenness centrality change (0.19), and distinct co- authorship communities (0.16) as top predictive features. The system achieves F1-score 0.838, ROC-AUC 0.89, and Silhouette Coefficient 0.65, enabling predictive identification of emerging leaders at least two semesters in advance.
Disclaimer: Curated by HT Syndication.
