MUMBAI, India, Sept. 22 -- Intellectual Property India has published a patent application (202641108816 A) filed by Manas Kumar Yogi; and Dr. A. S. N. Chakravarthy on September 10, 2026, for System For Decomposing And Routing An Analytics Query Across Heterogeneous Privacy-Computation Backends Based On Per-Attribute Privacy-Mechanism Flags.
Inventors include Manas Kumar Yogi; and Dr. A. S. N. Chakravarthy.
The application for the patent was published on September 18, 2026, under issue no. 38/2026.
Abstract: ABSTRACT System for Decomposing and Routing an Analytics Query Across Heterogeneous Privacy-Computation Backends Based on Per-Attribute Privacy- Mechanism Flags The present invention relates to a privacy-aware analytics query processing system configured to execute a single analytics query across heterogeneous privacy-computation backends based on per-attribute privacy-mechanism flags. The system receives an analytics query referencing a plurality of data attributes and retrieves a stored privacy-mechanism flag associated with each attribute. The flag identifies an appropriate privacy-computation technique, including homomorphic encryption, secure multi-party computation, or differential-privacy noise injection, according to the classification assigned to the respective attribute. The system automatically decomposes the received analytics query into a plurality of coordinated sub-queries according to the retrieved privacy-mechanism flags. The respective sub-queries are routed to corresponding privacy-computation backends, wherein the homomorphic-encryption backend performs computations on encrypted values, the secure-multi-party-computation backend performs joint computation using protected inputs, and the differential-privacy backend generates statistical results with calibrated noise. Attributes sharing a common mechanism may be grouped into a common sub-query where query semantics permit. Partial results generated by the heterogeneous privacy-computation backends are received and reconciled by a result-combination mechanism to produce a single consolidated analytics response. The system further supports dynamically updated privacy classifications, routing-decision caching, fallback routing, and cost-based selection for ambiguous or conflicting privacy-mechanism classifications. The proposed architecture enables heterogeneous privacy-preserving computation within a unified query-execution workflow without requiring the requesting application to independently manage the individual privacy-computation backends.
Disclaimer: Curated by HT Syndication.
