QBucket

Inspiration

We believe quick assessments are an effective way of learning. Most of the assessment platforms have limited questions and lack a wide range of topics. Inspired by the open-source culture, we developed a platform where people are free to contribute their ideas in the form of questions and discussions and gain knowledge in areas of their interests. We believe that this social assessment platform will revolutionalize learning and make it more fun.

What it does

1.) Promote an environment with endless opportunities for curious minds. 2.) Our system has two main functionalities, a.) Provide questions to the QBucket platform. b.) Take assessments at your own pace on a wide variety of topics of your choice.
3) Gives an opportunity to provide unique questions to the world.

How we built it

After brainstorming on this, we came up with a simple system design. We used HTML, CSS, and vanilla JS for the frontend, and for the backend, we explored Python, Flask framework, MySql, and MongoDB database. For topic detection, we used Bert, and to avoid duplicate questions we used Bi-LSTM.

Challenges we ran into

1) Lack of existing data sets. 2)Faced errors pertaining to Cuda while executing BERT.

Accomplishments that we're proud of

1) Created a dataset from scratch by collecting 400 plus data instances. 2) Developed a quiz portal to conduct quick assessments and roll out the performance for the quiz. 3) Implemented LSTM over the collected dataset and achieved an accuracy of 56%.

What we learned

1) Learnt data-preprocessing during the initial stage of the project. 2) Created a higher-level system design with scope for scalability. 3) Usage of jquery to modify the DOM elements. 4) We have learned to structure our thoughts for planning a business and understanding the opportunities for turning a vision into a reality.

What's next for QBucket

1) We plan to scale the project as a platform for hiring. 2.) Used as an internal assessment tool for organizations. 3.) We plan to extend a discussion forum to QBucket.

Domain Name: qbuckets.space

Built With

Submitted to

SacHacks III

Created by

I am responsible for creating the dataset for performing the category analysis. I used BERT and LSTM NLP models for text classification task. I helped in improving the performance of the model with tuning the hyper parameters of the model and increased the accuracy from 44% to 55%.

Shivaleela Ishwarappa Hubli
I am responsible for data set creation and implemented of NLP models such as LSTM for text classification. The goal was to achieve an accuracy of 80-90% for the LSTM model but under the limited time constraint I was able to achieve an accuracy of 55%.

Rhea Carmel Glen Rodrigues
I worked on developing Python Flask and MongDb implementation. I setup the entire backend architecture on AWS. Helped in designing the architecture and workflow.

Vamsi Krishna Velivela
Sri Sai Bhargav Koritala