MUMBAI, India, Oct. 26 -- Intellectual Property India has published a patent application (202541038318 A) filed by Ad Idem Technology Solutions LLP; and Poonamalle High Road, Kilpauk, Chennai- Tamilnadu, India Tamil Nadu India on April 22, 2025, for System And Method For Coherent, Structured Text Generation Using Semantic Similarity-Based Context Segment Arrangement.
Inventors include Adarsh Sosale; and Rohan K. George.
The application for the patent was published on October 23, 2026, under issue no. 43/2026.
Abstract: A system (100) and method for generating coherent structured text from a plurality of discrete textual context segments received from heterogeneous sources. A connection point determiner applies a plurality of analytical modalities - comprising semantic similarity analysis, syntactic parsing, coreference resolution with entity linking, and discourse structure analysis - to identify connection points between the segments and to place the segments on a two-dimensional coordinate space presented at a display device (112). A connector selector selects linguistic transitional connectors from a predefined set based on a determined semantic relationship between connected segments. An instruction set governs segment integration, coherence, contextual alignment, and style, and triggers query expansion for segments of low information density. A text-generating machine learning model (108) processes the linked segments with the instruction set, and a refinement module iteratively refines output against predefined quality criteria including coherence, factual accuracy, and stylistic adherence. Fig. 1
Disclaimer: Curated by HT Syndication.
