The problem QuizZen solves

This project is all about helping the students get a good understanding of the daily online lectures. QuizZen is an app where the teacher can create a room for uploading a quiz and the students present in that room will attempt the quiz. The teacher will not have to waste his time searching for questions on the quiz. He can simply upload a youtube link of a video and the quiz link will be sent to the students.

Challenges we ran into

We didn’t find any challenges in our frontend part, but in the ML part we faced some challenges due to which we had to narrow down our project. So, our aim in the starting was to generate MCQ quiz from youtube video. Since, youtube provided the transcript of any video, its difficult to apply NLP in the summary because it doesn’t contains any punctuation marks. So, we thought of resolving this error by making our project for Wikipedia, i.e. the teacher will only have to provide the topic in the frontend and he’ll be provided with the questions. But we were not able to generate good distractors(wrong answers) for the MCQ based questions, because our dataset was too small to train word2vec in our corpus. Therefore, we made fill in the blanks based questions. We were not able to merge these two components due to time constraint.