Inspiration
It's very overwhelming when moving to a new country where a different language is spoken. There are around 280,000 words in the English dictionary as of 2023. If someone has learned the language a little or is still learning they will forget certain words from time to time. This is where we thought of the idea of creating an application to help people remember words that are on the tip of their tongues.
What it does
termhint makes the user write a definition or a few words that relate to a word that they can't remember. The website will then showcase a list of words the user could be thinking of.
How we built it
The front-end of the website is made using:
- HTML
- Css
- Python (framework)
The actual functionality was made using:
- Python
- Tensorflow
- Pandas
- Numpy
Challenges we ran into
We couldn't get a generative AI like chatGPT to give a list of words the user could be thinking of accurately. So we had to go with plan B and train our own model that would recognize a few words based on definitions of words. Finding the right dataset for our project was a huge trouble for us as many datasets were too big for our model to handle, and the rest were too small to use. We ended up combining 3 different datasets which consisted of 12,000+ words and 100,000+ definitions.
Accomplishments that we're proud of
We are proud to get a working final product, we worked very hard to find datasets and learn to train models on them. Also, we are pretty happy to get a nice UI working with the Python files.
What we learned
This project taught us:
- How to preprocess sequence data to be used for an NLP model
- Having Python backend interaction with flask
- How to host our full-stack online
What's next for termhint
There are a lot of extra features we thought of that would make termhint more accessible and cooler! Allowing the user to talk to the app instead of typing, natively porting the app to mobile devices, including multiple languages, improving the accuracy and expanding the datasets
Log in or sign up for Devpost to join the conversation.