MUMBAI, India, May 1 -- Intellectual Property India has published a patent application (202611024010 A) filed by Dr. Arpita Basak; Dr. Deepa Sharma; Dr. Jaymala A. Patil; Dr. Suman Gulia; Ms. Pooja Sharma; Mr. Mayank Sharma; and Dr. Jyoti Malik, Ambala, Haryana, on Feb. 28, for 'system and method for automated classification and cataloging of library materials using artificial intelligence and machine learning.'
Inventor(s) include Dr. Arpita Basak; Dr. Deepa Sharma; Dr. Jaymala A. Patil; Dr. Suman Gulia; Ms. Pooja Sharma; Mr. Mayank Sharma; and Dr. Jyoti Malik.
The application for the patent was published on May 1, under issue no. 18/2026.
According to the abstract released by the Intellectual Property India: "The present invention discloses an AI-powered automated system and method, designated the ALMACC System (AI-based Library Material Automated Classification and Cataloging System), for the classification and cataloging of library materials including books, journals, theses, audiovisual items, and digital resources. The invention addresses the persistent challenges of manual cataloging in libraries, including inefficiency, high labor costs, inconsistency, inability to handle multilingual resources, and failure to scale with rapidly growing collections. The ALMACC System employs a multi-modal data ingestion module capable of receiving inputs via ISBN/ISSN lookup, scanned cover images, digitized full-text documents, structured metadata XML feeds, and manual entry. A dual-stream AI classification engine comprising a BERT-based Natural Language Processing (NLP) model and a Convolutional Neural Network (CNN) image classifier processes these inputs in parallel. Outputs from both streams are combined through a trainable ensemble fusion layer to produce unified, accurate classification assignments in accordance with the Dewey Decimal Classification (DDC) and Library of Congress Classification (LCC) schemes, accompanied by confidence scores. A catalog record generation module produces complete bibliographic records conforming to MARC 21, Dublin Core, MODS, and JSON-LD metadata standards. Records include DDC/LCC classification numbers, Library of Congress Subject Headings (LCSH), and all required bibliographic fields. The system integrates with major Integrated Library Systems (ILS) including Koha, Ex Libris Alma, and SirsiDynix Symphony via standard APIs, enabling seamless insertion of generated records into existing library databases. A Quality Assurance (QA) module validates generated records against authority files, applies deduplication algorithms to prevent duplicate catalog entries, and routes uncertain records to a human review queue with explainable AI-generated classification rationale. An automated reinforcement learning feedback loop captures librarian corrections and uses them to continuously fine-tune the AI models, ensuring that the system's accuracy improves progressively over time and adapts to institution-specific cataloging practices. A multilingual processing module supports automated cataloging in over 25 languages including Hindi, Sanskrit, Arabic, Chinese, French, German, Spanish, and Urdu, making the system broadly applicable across national libraries, academic institutions, public libraries, and archival repositories managing diverse and multilingual collections. Experimental evaluation at Maharishi Markandeshwar (Deemed to be) University Library demonstrated that the ALMACC System processes a batch of 500 mixed library items in approximately 4.2 hours with an average AI confidence score of 93.8%, achieving DDC number assignment accuracy of 94.7% and subject heading precision of 91.2%, compared to 16-21 working days required for equivalent manual cataloging by a team of three professional catalogers. The system achieves estimated cost savings of approximately 94% per cataloging batch, while improving catalog quality and accessibility. The ALMACC System is hosted on a secure, ISO 27001-compliant cloud infrastructure with role-based access control, data encryption at rest and in transit, and full audit trails, ensuring data security and regulatory compliance. A mobile application extends system accessibility to field librarians performing acquisitions away from desktop workstations. The invention constitutes a novel, inventive, and industrially applicable contribution to the fields of library and information science, artificial intelligence, and knowledge management technology."
Disclaimer: Curated by HT Syndication.
