MUMBAI, India, Sept. 14 -- Intellectual Property India has published a patent application (202641106601 A) filed by Kkr & Ksr Institute Of Technology And Sciences on September 04, 2026, for Adaptive Hybrid Framework For Real-Time Ad Click Fraud Detection.
Inventors include Mr. Prof. Repudi Ramesh; Mrs. Dr. Chittineni Aruna; Mr. B. Satyanarayana Reddy; Kolliboina Venkata Naga Yaswanth Kumar; Muthyala Ram Chandu; Mupparaju Harshavardhan; and Manam Simhadri.
The application for the patent was published on September 11, 2026, under issue no. 37/2026.
Abstract: ABSTRACT [0010] With the rapid growth of digital advertising and online platforms, a large volume of user interaction data is continuously generated through advertisement clicks and engagements. Many of these interactions are invalid or fraudulent, often generated by automated bots or repeated manual actions, leading to inaccurate campaign analytics and financial losses for advertisers. Detecting such fraudulent activities is a critical challenge, as traditional rule-based systems are unable to adapt to evolving fraud patterns and complex behavioral characteristics. This invention proposes an adaptive hybrid framework for real-time advertisement click fraud detection in digital advertising environments. The proposed system captures user interaction data and processes it using structured feature engineering techniques, including behavioral, temporal, and device-based attributes. A hybrid analytical approach is employed, where a random forest classifier evaluates individual click characteristics, and a long short-term memory network analyzes sequential interaction patterns to identify suspicious behavior. A key feature of the proposed system is its ability to classify interactions into legitimate users, repeated manual fraud, and automated bot activity. [0011] This is achieved through the integration of machine learning and deep learning models, enabling the system to detect both static and dynamic fraud patterns effectively. The framework also supports continuous monitoring and adaptive learning, allowing it to evolve with changing fraud strategies and maintain high detection accuracy. Experimental evaluation is conducted using simulated interaction datasets that represent real-world advertising scenarios with both genuine and fraudulent activities. The results demonstrate that the proposed system effectively identifies fraudulent interactions in real time while reducing misclassification rates. Overall, the proposed approach provides a reliable, scalable, and efficient solution for improving security, accuracy, and trustworthiness in digital advertising systems.
Disclaimer: Curated by HT Syndication.
