MUMBAI, India, July 11 -- Intellectual Property India has published a patent application (202541060941 A) filed by MLR Institute of Technology, Hyderabad, India, on June 26, for 'system/method for enhanced concept map generation from domain text using deep learning techniques.'
Inventor(s) include Mrs. S. Viharika; Mrs. J. Adilakshmi; Dr. Venkata Nagaraju Thatha; and Mr. B. Veerasekharreddy.
The application for the patent was published on July 11, under issue no. 28/2025.
According to the abstract released by the Intellectual Property India: "A concept map is a loose semantic knowledge representation graph that effectively organizes, represents, and visualizes knowledge present in a text. The concept map has to provide an overview of the document that is concise and effortless to understand. Representing and organizing the concepts along with their importance and with maximum information content is a vital aspect of the concept map. Concept maps have been constructed and used for many text mining applications, including summarization. In this work, machine learning algorithms and deep learning techniques are formulated to enhance the concept map and produce a concise and precise concept map from domain text. In this work, a new unsupervised graph-based algorithm capable of directly extracting domain phrases precisely without the use of domain resources is designed. A biased random walk-based overlapping small communities detection algorithm was used on graph-of-words of the domain corpus to identify the proper set of concepts that forms the domain vocabulary. The proposition extraction process was improved by the neural open information extraction method with the tree-based convolution neural network. The use of advanced deep learning techniques alleviated the need for feature engineering for proposition extraction. In addition, precise argument extraction was achieved with dependency tree processing and the domain vocabulary generated using unsupervised graph-based key phrase extraction. Finally concept importance and topic identification needed for the enhanced concept map is obtained by sub-topic detection based on partitional clustering using K-means and HDP."
Disclaimer: Curated by HT Syndication.
