MUMBAI, India, Sept. 14 -- Intellectual Property India has published a patent application (202641108088 A) filed by Shaheena Kv; Anila R. Nambiar; C M Sulaikha; Reshmi V; and Prapitha Gopi K on September 08, 2026, for An Ai- And Machine-Learning-Based System And Method For Detection Of Fake Profiles In Social Networks Using Random Forest Classification And Natural Language Processing.
Inventors include Shaheena Kv; Anila R. Nambiar; C M Sulaikha; Reshmi V; and Prapitha Gopi K.
The application for the patent was published on September 11, 2026, under issue no. 37/2026.
Abstract: [044] A computer-implemented system and method are disclosed for detecting fake profiles in social networks by multi-modal artificial-intelligence and machine- learning analysis. Permitted account data are normalized into a canonical profile representation and processed to derive NLP features from profile and activity text, behavioural and temporal features, relationship-graph features, cross-modal consistency features, and evidence-quality indicators. The features are fused into a unified feature vector and evaluated by a Random Forest ensemble to obtain a fake-profile probability and a tree-disagreement measure. A calibration and risk-aggregation engine generates a confidence-aware risk score using the ensemble output, evidence quality, and optionally a bounded cohort-risk value derived from content similarity, temporal synchronization, target overlap, or graph similarity. A two-pass embodiment introduces bounded cohort feedback to improve detection of coordinated deceptive accounts while limiting cascading false positives. An explanation generator identifies principal contributing factors and supports threshold-based flagging, human review, or non-action. The architecture thereby provides platform-adaptive, explainable and sparse-data-aware detection of fake, impersonating, automated, or coordinated profiles. Accompanied Drawing [FIGS. 1-2].
Disclaimer: Curated by HT Syndication.
