EZUX Redesign: NFT Dashboard and Metrics

'+++ Measuring the quality of NFT creators and NFT artworks+++

**Component-1: A comprehensive creator dashboard or a public creator profile page. **

• [ ] Creators are incentivized to fill out the details, e.g., prior experience in NFTs and other content creation, and social media accounts. Interested buyers or collectors can then check the creator profile before purchasing the NFTs.

• [ ] System provides public and on-chain data: the number of followers on social media (i.e. Twitter) and in NFT marketplaces, record of interpersonal communication in the marketplace, history of prior spam incidents, and NFT ownership record integrated into the profile.

Impact

• [ ] With such a public profile system, NFT creators can also look for potential collaborations and see the updates of which NFT art categori are in trend. New artist can know what area they would like to put some more effort on
• [ ] Machine learning models could be trained, taking the information mentioned above as the input, to classify legitimate vs. scam NFTs. Providing the model predictions to users can help them make more informed buying decisions, and reduce the chance of falling for scams.
• [ ] Investment companies can have more information for trading. Facilitate trust and reduce ambiguity

Component-2: Visual hashing with state of art computer vision to compute visual similarity of two NFTs

• [ ] Apply existing computer vision techniques (e.g., visual hashing) to automatically compute visual similarity of two NFTs. This assumes that there is a database of existing NFTs, which is not far-fetched because virtually all NFTs should be public and on-chain.

Impact

• [ ] When visual hashing detects highly similar NFTs, it can alert community members

The project idea came from my academic user study of interviewing NFT creators and buyers. Insights and evidence of qualitative studies are the motivation for this project. I mainly reflect on the possible design implication and finally redesigning the Dashboard of NFT creators and visual hashing taking OpenSea as a use case.

*** First used Google drawing for the ideation of concepts for the dashboard and visual hashing.

*** React JS, HTML, CSS to build UI front-end design from component-1: creator Dashboard or a public creator profile page

*** Created a small-scale simulated data to implement the visual hashing to compute the visual similarity of two NFTs. Mainly to calculate the level of confidence for similarity, subsequently identify potential copy NFTs and scams. I have used state of art CV algorithm from prior literature "Few-shot learning"