keyboard_arrow_up
Efficiently Navigating Information Overload: Developing a Machine-Learning-Based Application for Summarizing Academic Videos and Extracting Key Topics

Authors

Tianyang Wang1 and Aleksandr Smolin2, 1USA, 2California State Polytechnic University, USA

Abstract

The internet is a vast treasure trove of information on any subject that has allowed students and scientists alikeunprecedented access to the world's collective knowledge [1]. Unfortunately, a lot of it is buried in hours longseminars, talks and other videos on video sharing websites like YouTube [2]. When researching a topic for anacademic essay or paper, one might rightly be shocked by the massive time investment required to dig uptherelevant data or citations from these monolithic recordings [3]. This paper develops an application that aims toapply new machine-learning-based transcription and keyword extraction methods to cut these videos into small, digestible chunks, which are labeled with their most important topics in order to allow us only to have to manuallyanalyze the parts of the video that are relevant to our research without losing valuable context details [4]. Weapplied our program to instructional videos on YouTube, in order to test how well we can rearrange videodocuments for a more convenient view of its contents and conducted a qualitative evaluation of the approach. Theresults show that the application works as expected to provide video clips titled with its central topic for the user todownload in a reasonable amount of time via a simple browser-based extension [5]

Keywords

Python Flask, Firebase, Assembly AI, Google Chrome

Full Text  Volume 13, Number 9