MUMBAI, India, Aug. 12 -- Intellectual Property India has published a patent application (202641087574 A) filed by S. Kirthica; R. Shathanaa; I. Dharshini; and P. Shanthi on July 17, 2026, for Multi-Lingual Multimodal Legal Intelligence Framework.
Inventors include S. Kirthica; R. Shathanaa; I. Dharshini; and P. Shanthi.
The application for the patent was published on August 07, 2026, under issue no. 32/2026.
Abstract: A computer-implemented system is provided for enabling bilingual legal research using a multimodal artificial intelligence framework. The system is configured to receive input in multiple modalities, including text, speech, and image data. An automatic speech recognition module converts spoken input into textual fonn to support users with limited typing capability or digital literacy. The system includes a bilingual interface supporting Tamil and English, wherein user queries may be submitted in either language. A language normalization module translates non-English queries into English, such that all downstream processing is performed in a single canonical language. A retrieval-augmented generation architecture is employed in which a primary language model performs intent interpretation on the normalized English query. The processed query is used to conduct semantic similarity search over a vector database comprising legal statutes and judicial decisions stored exclusively in English. By maintaining the legal knowledge base in a single language, the system avoids data redundancy and ensures consistency in legal interpretation. Retrieved results are processed by a domain-adapted generative model within a reasoning engine to produce context-aware legal outputs. The generated response may be translated into Tamil or retained in English, depending on user preference.
Disclaimer: Curated by HT Syndication.
