1 of 21

RSVP_to_finale

Ritwiz Sinha | Saksham Jha | V.S Subhang | Prem Bhawnani

COVID 19 AND TECHNOLOGY

IIT PATNA

2 of 21

Lecture Forager

Optimizing mindless scouring through college lectures

3 of 21

Demo

Diving in

The Product

Product Demo

Technical Details

Motivation,problem and its solution

4 of 21

The Product

Problem Solved, Novelty and Use Cases

5 of 21

Motivation For the Idea

In this pandemic most of the college students have had online tests, online classes and even online labs whenever possible.
The format for discourse has primarily been online lectures typically ranging from 1 - 2 hrs as what happened in normal offline classes.
Note making has been severely affected due to this. In online classes as exams approach the students start viewing the lectures haphazardly thinking to retain this information.
Searching for a small topic explained by the professor in an array of unintuitively named video files is not an easy task.

6 of 21

PROBLEM

STATEMENT

There is a lot of back and forth while studying through online lectures if you want to study a particular topic then you have to search through all the videos where that topic was mentioned to create notes or even understand the topic. This wastes a lot of time while studying and distracts the focus of the student.

7 of 21

SOLUTION

Our team created a web app search engine for college lectures where one can search for a keyword and get results of videos in which that keyword was mentioned by the professor during the lecture.

Instead of using video metadata for indexing in search engine we use the keywords in the transcript of the video

The application also redirects the user to the timestamp around which the word was used when playing the video.

8 of 21

SOLUTION

When the video is playing we also provide all of the important keywords that were used in the video along with their timestamps.

Drastically reduces search time for huge college lectures and helps us learn topics in short and concise manner.

11 of 21

Technologies and Frameworks Used

AWS Services

AWS Transcribe API
AWS Lambda
AWS EC2
AWS S3

Redis with task queues
Node with express
Webpack
React
Typescript/Javascript
Docker
Postgresql

12 of 21

Demo

Product Demo

13 of 21

Diving In

Problem Solved, Novelty and Use Cases

14 of 21

Architecture

Understanding the bits and pieces

16 of 21

How it helps

No binary searching through video files and videos
Optimizing selective study in recorded lectures
Minimizes extra effort made while labelling video parts for easy recognition
Get to know about other topics covered in the video without watching video
Can be extended to summarise the lectures and create automated note making systems.

17 of 21

NOVELTY

Usage of transcript is very limited on education based applications like Udemy and Coursera, which just allows you to go through it or reach a particular part of the video.

In our idea we have extended the use of transcripts.

Along with presenting the transcript to the user we extract keywords from it and create a search engine which enables people to search for all the instances of videos where a particular keyword is present.

18 of 21

Technical Improvements

Create fully serverless backend
Improve the search engine capacity and efficiency by using services like Elasticsearch.
Using AWS Comprehend for getting keywords and topics
Instead of manual uploads add plugins for direct upload from Google Meet and Zoom.

19 of 21

Expanding the product

Self Hosted solution for colleges

Collaboration platform

Provide a SaaS platform to colleges and educational institutions to integrate in their systems.

Integration with hundreds of videos at platforms like NPTEL