Botiful

Inspiration 🤖

We wanted to make an NLP-powered bot that can aid a wide range of technical professions from doctors to lawyers. Many highly trained careers require rote memorization of tons of technical jargon and even research. An aid that is always available would be extremely stress-relieving for them.

What it does 🤖

Basically what our inspiration states, but the chatbot can perform a variety of language related tasks, including being a technical dictionary, search Wikipedia, and find important terms in a text.

How we built it 🤖

The bulk of the functionality was coded in Python. 🐍 The functionalities of the backend include using Google Cloud Vision ☁️ to perform OCR text recognition in images. 🔍 Another major helper was the transformers library from HuggingFace 🤗 that uses state-of-the-art BERT model to support pipelines for summarization, named entity recognition, Q/A, sentiment analysis, and more. 💬 Finally, DataStax Astra helped the bot to store chat as a history log, which can be searched by keywords.

Challenges we ran into 🤖

Differing timezones and skills marked the main challenge we faced.

Accomplishments that we're proud of 🤖

Making the backend work with BERT model was the biggest accomplishment.

What we learned 🤖

see above

What's next for RoboHacks 2021 🤖

Finishing up the aesthetics of the website and possibly making a console app version.

Built With

• cassandra
• css
• google-cloud
• html
• javascript
• linode
• python
• pytorch
• tensorflow