Neura

Text sentimental analysis
Video sentimental analysis
Texts are summarized and questions are generated using AI
Audio sentimental analysis
Result of sentiment analysis
Neura Interview platform

makeUC

Video Link

Inspiration

Neurodiversity is a form of diversity that has numerous benefits in the workplace. People with neurocognitive disabilities have unique talents, views, and skills that can be very useful in a variety of workplaces. Employers are increasingly acknowledging these advantages and developing hiring initiatives aimed at attracting neurodiverse staff. While these initiatives are more popular in larger firms, they have proven to be useful for businesses of all sizes across a wide range of industries. Hiring neurodiverse employees can give firms a competitive advantage that results in concrete financial and workplace culture benefits.

What it does

Neura is a revolutionary learning platform for neurodivergent people which furthermore provides a conducive interview experience. This platform employs artificial intelligence to summarises texts, highlight salient ideas, and generate questions. Moreover, It is also an interviewing platform where employers can conduct three different forms of interviews namely, text, audio, and video interview. All three sorts evaluate the textual, vocal, and facial emotional score respectively. This score can be used to ask questions that would make an interviewee comfortable. Additionally, the interviewee can see his sentimental analysis score to improve his performance further.

How we built it

Neura is built on a Flask micro-framework. It uses open-source pre-trained Machine Learning models. Natural Language Processing is used to summarise the text and to generate the questions. HTML, particle.js, CSS is used in frontend technologies. For text analysis, custom NLP like Tokenization of the document, cleaning, and standardization of formulations using regular expressions, deletion of the punctuation, etc is used and 3 layer LSTM is then used in order to leverage the sequential nature of natural language. For the audio interview, Time Distributed Convolutional Neural Network is used which uses used 4 Local Feature Learning Blocks (LFLBs) and the output of each of these convolutional networks will be fed into a recurrent neural network composed of 2 cells LSTM. XCeption model is used to evaluate the sentiment in the video-based interview, the pixels are being activated differently depending on the emotion being labeled.

Challenges we ran into

Time constraint was the first challenge. Integration of all the models in the project was a major challenge. Another challenge we faced was the accuracy of the ML models.

Accomplishments that we're proud of

We brought the community closer by implementing various Machine Learning models. We helped neurodivergent people get an equal opportunity by improving their knowledge of lengthy concepts which they might find difficult to focus on. The interview process in Neura will make the neurodivergent audience much more confident and employers can understand them better. This will create an inclusive environment. We are proud of integrating and using many models in the platform in less than 24 hours!

What we learned

We learned about various models of NLP and emotion sentiment analysis. We learned about the particle.js and we learned about the integration of various models in one portal.

What's next for Neura

A new feature of "ask an expert" will be added in the future where participants can ask the generated questions to the expert. Neura will become a fully voice-controlled website in the future.