MUMBAI, India, June 30 -- Intellectual Property India has published a patent application (202641076188 A) filed by Sr University on June 19, 2026, for A Resolving Lexical And Structural Ambiguities In English-Hindi Translation Systems.
Inventors include Priyanka Suram; and Dr. Pramoda Patro.
The application for the patent was published on June 26, 2026, under issue no. 26/2026.
Abstract: A RESOLVING LEXICAL AND STRUCTURAL AMBIGUITIES IN ENGLISH-HINDI TRANSLATION SYSTEMS An enhanced English-Hindi translation system and method are disclosed for resolving lexical and structural ambiguities via a multi-level context-aware pipeline. The system comprises a lexical disambiguation module that utilizes bilingual dictionaries, WordNet synset mapping, and contextual BERT or mBERT embeddings to select accurate word meanings based on surrounding context. A structural disambiguation module incorporates dependency and constituency parsing alongside semantic role labeling (SRL) to align English sentence patterns with syntactically equivalent Hindi structures, effectively resolving ambiguities from relative clauses and word-order variations. The system outputs are integrated as ambiguity flags into a transformer-based neural translation engine, such as mBART or MarianMT, which is trained on an augmented dataset with ambiguity tags to generate fluent, accurate Hindi text. A continuous feedback loop evaluations model performance against an annotated benchmark corpus to facilitate adaptive learning and dynamic rule refinement.
Disclaimer: Curated by HT Syndication.
