MUMBAI, India, Jan. 9 -- Intellectual Property India has published a patent application (202521108354 A) filed by Alpesh Prakash Sonar; and Divyani Alpesh Sonar, Thane, Maharashtra, on Nov. 8, 2025, for 'integrated data annotation framework for detecting missing labels in machine learning workflows.'
Inventor(s) include Alpesh Prakash Sonar; and Divyani Alpesh Sonar.
The application for the patent was published on Dec. 12, under issue no. 50/2025.
According to the abstract released by the Intellectual Property India: "An integrated annotation assistant system (100) automates identification and completion of missing labels within a no-code machine learning workflow. The system (100) includes a receiver (104) of a server (106) configured to receive data from a user device (102), the data comprising text, image, or tabular data. A processor (108) retrieves and processes the data to detect missing or inconsistent labels using n-gram pattern detection, metadata analysis, or statistical correlation checks. The processor (108) employs trained machine learning models-a transformer network for text, a convolutional neural network for images, and a decision-tree classifier for tabular data to generate label suggestions. Suggested labels are displayed on a display (110) for user confirmation or correction. Verified labels are validated against stored ontologies and schema standards such as COCO or CoNLL, and the dataset is automatically updated, enabling real-time, automated, and schema-compliant annotation within the same machine learning workflow."
Disclaimer: Curated by HT Syndication.
