LectureLM: Intelligent Lecture Understanding and Interactive Learning Model

We presented LectureLM, a Python-Streamlit application that transforms YouTube lecture videos into structured, adaptive, and interactive study sessions. The system addresses a genuine gap in the current landscape of AI-powered educational tools: it does not merely summarize, but learns from and adapts to the learner’s performance. Key contributions include the four-layer architecture, the session Memory module, the RAG-grounded Q&A interface, and the adaptive quiz engine. Remaining work includes multimodal inputs, persistent user profiles, and a formal user study with statistical analysis.

[1]  F. Retkowski et al., “End-to-End Neural ASR with Fluency Error Detection,” arXiv:2504.08024, 2025. [2]  L. Benedetto et al., “AI-Generated Summaries for Educational Video Content: Challenges and Opportunities,” J. Educational Technology, vol. 18, no. 2, pp. 45–62, 2022. [3]  P. Ilampirayan et al., “Automatic Lecture Note Generation Using NLP,” Global Int. J. Emerging Technologies (GIJET), vol. 11, no. 2, pp. 35–48, 2021. [4]  S. Gupta and R. Sen, “Abstractive Summarization of Educational Videos Using Pretrained Transformers,” Int. J. for Multidisciplinary Research (IJFMR), vol. 8, Nov.–Dec. 2023. [5]  J. Holt et al., “BERT-Based Topic Clustering for Lecture Content Analysis,” arXiv:1906.04165, 2022. [6]  A. Bhargavi and V. Subha Rao, “Whisper-Powered Personalized Learning: GPT Models in Adaptive Study Environments,” arXiv:2410.04449, Oct. 2024.