SOURCE-LINKED INTELLIGENCE

KnowVis: Knowledge-Centric Visual Summarization for Video Lectures

arXiv · AI, language, vision and robotics · article · Sep 3, 2026 · UTC

Video lectures are valuable educational resources, but their dense and lengthy formats often overwhelm novice learners. This difficulty stems from a fundamental pedagogical mismatch: while videos deliver transient information linearly, human learning requires constructing interconnected cognitive networks, a task that induces severe cognitive overload for novice learners lacking prior domain knowledge. Existing video summarization methods fail to resolve this mismatch, as they primarily produce text-heavy, linear condensations that still demand high cognitive effort. To bridge this gap, we pro

Read original source ↗ Open in workspace

recordType

paper

region

Global