Many times we might have come across a situation where we lose concentration while watching a free online lecture video. A short survey that we took during the past few hours supports this fact.
What it does
To solve this unaddressed problem, we have come up with an artificial intelligence-powered solution Ailee, which asks questions to the viewer and makes the experience more engaging. When the user answers wrongly Ailee would recommend them to watch the part of the video that has the necessary imformation.
How we built it
A lecture contains three key components, speech, visual(input from the board) and transcripts, from which we have to understand the context to provide a good set of questions. Used text detection, HTR techniques by using tensor-flow to get the text from the board. Used an online tool voice-notepad to get the text out of the speech. Employed NLP techniques using nltk to understand the context and form questions.
Challenges we ran into
Making a contextual aware system to ask meaningful questions is where had challenges
Accomplishments that we're proud of
We are leveraging the existing educational resources to provide a cost-effective solution. Employed AI to overcome the overhead of manual laboring intensive task of creating question sets.
What's next for Ailee
We are planning to further expand the capabilities of Ailee by implementing recommendations of other online resources which will help the struggling viewer to understand concepts from different resources.