As the world around us struggles to cope with the online and remote methods of work and education, we often find ourselves looking at the screen and trying to concentrate on a meeting/class.
The problem with an online meeting/class is not being able to deliver a fulfilling and engaging interaction with everyone. While interacting in person we are continuously giving out signals via our expression and sounds as feedback to the speaker, which is mostly missing in the online meetings. So, we can dare to say that an online class becomes analogous to an audiobook.
The solution we propose is simple and here are the steps to realize it.
- We collect the transcript of the meeting and send it to our API in realtime.
- The API will read the text and create
fill in the blanks
andtrue/false
questions based on the text. - The questions are sent to participants of the meeting via a mobile app.
- The participants can answer the question.
- All participants and hosts can view the results of the test in realtime ( while the meeting is in progress ) on a leaderboard.
The technique to create questions from a text is one of the very popular subjects of AI research. Big players like Intel and Google have made substantial progress in this area. We are using simple ML techniques, some of the open-source work, and some regex to create the API.