High-Level Project Summary
We have created a web application having an artistic frontend and a robust backend of a search engine that helps search through NASA’s database, including images and research information.
Link to Final Project
Link to Project "Demo"
Detailed Project Description
Backend
backend.py file consists of 2 ML models based on KeyBERT and BERT.
Initially, the model extracts essential keywords from the user-entered string using a pre-trained customized BERT model, which returns a list of keywords. Then, the obtained keywords are processed for lexical simplification such that complex terms like ‘lunar’ are converted to ‘moon’ for an easily accessible and better search.
So basically, we improvised the previously implemented keyword-based model in the https://images.nasa.gov/ website using ML for a more precise search mechanism.
The ML model returns a list of keywords that are then sent to the API, which produces a JSON array of dictionaries with information about the images, their descriptions, and other related research information.
Frontend
The homepage.html, homepage.css, and homepage.js files are for the frontend.
The frontend has been made artistic and dynamic with the help of animations for an immersive experience.
The project is 80% complete. Due to the limited time, the only thing left is connecting the backend to the frontend, which includes sending the user-entered string to the backend and then displaying the research data along with images on the frontend.
Space Agency Data
After processing the user-entered string, we used the NASA images API to search through the images using the advanced keywords provided by our ML model. We have also used the NASA logo on our search page.
Hackathon Journey
Our team had a lot of fun throughout the event and learned a lot of new stuff related to HTML, CSS, JavaScript, and Machine Learning to tackle various challenges we faced while working on the project. We collectively thank NASA for such a fantastic event.
References
We have used a map of the moon and landscape clipart on the search page.
Bert has been used for creating ML models for the backend.
StackOverflow has been used to resolve some coding challenges we faced while working on the project.
Tags
#software, #art, #artistic, #ML, #machinelearning, #search, #frontend, #backend, #artistic, #bert, #mlmodel, #