MUMBAI, India, June 22 -- Intellectual Property India has published a patent application (202511040252 A) filed by Indian Institute Of Technology Jodhpur on April 25, 2025, for Heterogeneous Multilayer Density-Infused Entropy-Modularity Optimization For Unsupervised Social Event Detection.
Inventors include Suman Kundu; Shraban Kumar Chatterjee; and Shubham Gupta.
The application for the patent was published on June 12, 2026, under issue no. 24/2026.
Abstract: The present disclosure provides a system (100) for unsupervised social event detection from social media data streams, including a processor (102) configured to receive social media messages and construct a heterogeneous multilayer graph (104), a text encoder module (112) configured to generate feature representations, a community detection module (114) configured to perform density-based community detection using an mCOMM algorithm across multiple layers including a word- based interaction layer (106), an entity-based interaction layer (108), and a mention-based interaction layer (110), a community sketcher module (120) including a forest data structure (122) and sparse triangular matrix (124), a graphical projector module (118) configured to generate projection graphs (132), a graph encoder module (116) implementing a graph convolutional network encoder with residual skip connections, and an optimization module configured to perform combined entropy- modularity optimization through a loss function integrating entropy loss and modularity loss. (Figure. 1)
Disclaimer: Curated by HT Syndication.
