MUMBAI, India, Sept. 14 -- Intellectual Property India has published a patent application (202641094837 A) filed by Dhanalakshmi Srinivasan College Of Engineering And Technology on August 05, 2026, for Ai-Powered Smart Interview Assistant With Real-Time Feedback And Score Prediction.
Inventors include Dr. Umma Habiba Ph. D; Yogalakshini M; Jeni Pragashini J; and Kayathri A.
The application for the patent was published on September 11, 2026, under issue no. 37/2026.
Abstract: ABSTRACT OF THE INVENTION AI-POWERED SMART INTERVIEW ASSISTANT WITH REAL-TIME FEEDBACK AND SCORE PREDICTION The present invention relates to an Artificial Intelligence (AD-based intelligent interview assessment system and method for automating candidate evaluation through the integration of Machine Learning (ML), Deep Learning (DL), Natural Language Processing (NLP), Generative Artificial Intelligence (GenAI), Large Language Models (LLMs), Conversational Al, Speech Recognition, and Predictive Analytics. The system comprises a Data Collection Module, Data Preprocessing Module, Feature Engineering Module, Feature Fusion Module, Interview Performance Evaluation Module, Feedback and Score Prediction Module, Visualization and Dashboard Module, Report Generation Module, Security and Validation Module, and Database Management Module for performing end-to-end interview assessment. The invention employs the Qwen Large Language Model (LLM) through the Hugging Face Transformers framework to generate adaptive, role-specific interview questions and enable intelligent chatbot interactions. Candidate responses are acquired in both text and voice formats, wherein voice inputs are converted into textual representations using Speech Recognition and Speech-to-Text (STT) technology. The collected data undergoes preprocessing operations including text normalization, tokenization, and feature extraction before being transformed into contextual representations. The system utilizes DistilBERT for contextual sentiment analysis and Sentence Transformers (all- MiniLML6- v2) for semantic similarity analysis and text embedding to evaluate the relevance, quality, and contextual understanding of candidate responses. Semantic, linguistic, contextual, speech-derived, and candidatespecific features are fused into a unified feature vector, which is processed using Random Forest Regression, implemented through Scikit-learn, to predict interview performance scores and generate personalized analytical feedback. The invention further provides an interactive web-based dashboard for real-time visualization of interview analytics, candidate rankings, semantic and sentiment scores, performance metrics, and downloadable assessment reports while incorporating security, validation, and database management mechanisms to ensure reliable and secure operation. The proposed system delivers an objective, scalable, and data-driven interview assessment framework that minimizes interviewer bias, enhances evaluation consistency, improves recruitment efficiency, and supports recruitment organizations, educational institutions, corporate enterprises, and online assessment platforms.
Disclaimer: Curated by HT Syndication.
