Awards & Nominations
Data Without Limits has received the following awards and nominations. Way to go!
Global Nominee
Data Without Limits
High-Level Project Summary
The vast size and complexity of NASA’s public database(images and datasets) necessitates a public interface, able to be utilized not just by experts in the sciences, but also by everyone in pursuit of a wide variety of goals. Data Without Limits is a website designed to enable exactly this. Behind the simple, accommodating interface, Data Without Limits employs ML/AI software which provides intuitive access to NASA’s public database of datasets and images for individuals at any level of expertise. This software also enables modification of database images using Artificial Intelligence for creative expression, providing a means for creating and sharing art from the website.
Link to Final Project
Link to Project "Demo"
Detailed Project Description
A critical element surrounding the knowledge which science produces is its ability to be communicated. Over 10,000 datasets and images collected from numerous NASA missions are publicly available and though scientists with experience in research might be able to navigate their way through any database, for most of the public, an easily accessible, yet detailed database is required for this knowledge to be used to its full benefit. Essential to this database democratization is the provision of tools which allow the unique ways in which the general public can use this knowledge, whether that be in the pursuit of greater scientific understanding or as a means of creative self-expression through art.
Data Without Limits provides a digital interface that enables not only comprehensive access to the full breadth of NASA’s public database for individuals at any skill level, but also the ability to further their use of the data using built-in tools for deepening knowledge and creative expression. This is achieved using a website interface which strategically integrates ML/AI into a number of processes in data access which accentuates the user experience.
The user experience begins at a multi-modal search engine with access to 22,500 datasets and 10,000 images in NASA’s public database. This search engine uses ML/AI both to personalize input method to accommodate different forms of communication and to tailor search results according to those inputs. In addition to standard text input, this search engine allows for sketch input through the use of Stable Diffusion AI, allowing for a more visually inspired approach to data access. From this diversity of input, search results are provided through Algolia AI and AI Synonym Search, which provide search results that present users with a selection of datasets and images ranked by relevance according to the user’s input. This also accounts for inexact inputs such as basic misspellings or use of narrow descriptors, which can be successfully parsed and analyzed by the AI.
Once a user has selected the relevant data, they have a brief summary and tools for data use at their fingertips on a dedicated webpage for the particular data point. For a dataset, integration of OpenAI GPT-3 software into the interface allows users of any skill level to ask questions to the AI in a text field below the summary. This feature can be used to ask a host of questions about the data set, ranging in complexity from general inquiries about basic concepts to exact figures of complex data. This versatility provides considerable access to NASA’s public database for both specialists and non-specialists alike. It is capable of quickly retrieving highly specific data, while also democratizing otherwise inaccessible specialized knowledge. For an image, editing tools implementing Stable Diffusion AI provides creatively minded users to edit photos they find with the use of AI, transforming datapoints into art. They can also directly share the art they create with an audience on a number of social media platforms from the website. This promotes a use of scientific data that takes it beyond the sphere of technical exclusivity and offers a wider public use that can serve as a unique means of creativity, in doing so working to popularize science.
As a team, we had ambitions to move beyond English-language keyboard input and additional sketch input for images particularly. To further democratize NASA’s public database, we aimed to add multilingual support for the entire site, allowing many of the 6.3 billion non-English speakers worldwide to access the entirety of this knowledge as well. In addition to this, we aimed to add multilingual speech-to-text input as another way of diversifying types of input. However, due to time constraints, we were not able to add these features.
Space Agency Data
As our project revolves around creating an accessible interface for NASA’s public database, we sourced metadata from 32,500 datapoints in NASA’s public database through which our interface could draw data. These datapoints were comprised of 22,500 datasets from nasa.data.gov sourced using Postman and 10,000 images from images.nasa.gov sourced using JSON files provided by NASA.
Hackathon Journey
Prior to this hackathon, we never thought a two-day project could have the magnitude to be considered a journey, but in light of the intensity of ups and downs we have faced as a team and what we have accomplished, we certainly think so now. This journey can best be described as one of constant adaptation as we faced and overcame the various challenges associated with taking our vision for Data Without Limits from the drawing board to a real, functional interface/AI tool that makes NASA’s data creatively accessible to millions. However, none of this would be possible were it not for the support and hard work of the organizers of this year’s Space App Challenge and to them, therefore, we would like to extend our thanks.
References
Databases:
images.nasa.gov
data.nasa.gov
Software:
https://www.postman.com/
huggingface.com
https://stability.ai
https://openai.com/api/
https://colab.research.google.com/
Tags
#data #art #public #AI #knowledge #communication #input #sketch