Similarity Service in Marketplace for NFT Metadata

joe-p commented 2 years ago

Prize Bounty

1st Place - 10,000 ALGO 2nd Place - 7,500 ALGO 3rd place - 5,000 ALGO

Challenge Description

Gaming NFTs are difficult to price correctly because their value is heavily weighted towards what they offer in-game and less correlated to common data like name, image or rank, AlgoSeas PIRATEs, for example, have 4 in-game stats stored on the NFT as ARC69 metadata: combat, constitution, plunder, and luck. Sales data show that these in-game stats dominate above all else when it comes to valuing a pirate, with combat and constitution being the two most important stats. PIRATEs have been sold anywhere from 1-1,000 Algo. With over 15,000 PIRATEs in circulation and hundreds of pirates listed and sold it can be difficult for buyers and sellers, especially newcomers, to know what the fair value of a pirate is. We need a similarity service that accepts a single NFT and returns similar NFTs that have been recently valued by others (active listings or recent sales) based on weighted property data. The service has several interesting challenges:

Pirates always change Pirate stats go up and down as pirates level-up, fight, die, and are revived. Keeping track of changing pirate data is important. New pirates are also minted daily so it’s an ever evolving dataset. Some NFT collections don’t change but we expect lots of gaming NFTs and future ARC19 NFTs to have high mutability.
Tracking current listings and sales - Data is all available via web3 to do it from the blockchain or you can use provided web2 APIs to get this data (see resources section for details).
Listings vs Sales - A similar pirate that Active listings and sales can both be used for finding similarities. For example the service can find: this NFT is similar to NFT “A” that was sold ~3 days ago for X Algo, and is also similar to NFT “B” that is currently listed for Y Algo.
Weighted Property Data - How does the service decide which properties are most meaningful? You can come up with your own heuristic or you could analyze bulk sales data to try and compute weights.

All of the data needed for this challenge re tracking NFTs, listings, and sales is available via web3 or one can use the AlgoSeas APIs to get this data (see Resources section for details)

The challenge is open-ended but the service should effectively answer this question: “Given an NFT what are the closest N active listings and what are the closest N recent sales for this NFT?”

At minimum the response needs to contain the asset IDs that are most similar but it could also provide other data like the algo amount, date it occurred, whether it is a listing or a sale, a “similarity score” (how close your service thinks these two NFTs are to each other), etc.

Submission Requirements

A valid submission will include the source code via a Github repository linked to your Gitcoin submission, and a recorded demo, working CLI, or live API that can be called (for determining the latency criteria). We prefer the service to be written in TypeScript but will accept solutions in JavaScript or Python. This service will be folded into the AlgoSeas platform and maintained by the development team. We do not expect any frontend development for the challenge but if you want to build anything for demo purposes and to be usable by the team we prefer React. It is expected that most bounty challenges will be won by fully functioning submissions, however, in certain cases, where there is a higher degree of difficulty, projects and payouts will be assessed on progress made The project submission deadline is 17:00 EST on Wednesday the 2nd of November.

Judging Criteria

There will be a community vote that will commence shortly after the project submission deadline; this will close at 17:00 EST on Monday the 7th of November The community vote will primarily be for voting on NFT similarity and will be hosted in the AlgoSeas Discord. Two representatives from AlgoSeas and two other judges (by invitation) will also assess each project submission by 17:00 EST on Monday the 14th of November. The following criteria will be used to assess projects:

FUNCTIONAL

Does it work when run against any pirate AssetID? Are the similar results provided reasonable? How do your results for a specific assetID compare to those of other submissions? Does it work when Pirate metadata is updated or when new Pirates are minted?

LATENCY

How long does it take to get results from your service? Latency will be determined by running a series of requests and taking the p90 latency.

EXTENSIBILITY

How generic is your solution, could it be easily applied to other NFT collections? Some examples of collections that it can be applied to include “Alvatars: Tile Fighters” or Alchemon.

WOW FACTOR

Hackers always do surprising things. Thought of something that would be really cool to have that isn’t listed here? Go for it and you can earn bonus points from this category!

Terms

All submitted code will be considered usable by the AlgoSeas team, regardless of whether or not the participants are chosen to receive a prize bounty. Additionally, refer to the Participation Agreement for full details. Winner Announcement Date Winners will be announced on Sunday the 20th of November, with payouts to be completed no more than one week thereafter. Ecosystem Funding may be issued at the discretion of the Algorand Foundation to retrospectively advance high potential projects. Resources AlgoSeas IndexerAPI - https://algoexplorer.io/api-dev/indexer-v2 AlgoSeas Discord - https://discord.gg/HKQAGXKJc3 AlgoSeas Discord Sales Channel - https://discord.gg/pf898W2U2j Pirate NFT Creator address (NEX7N2...) AlgoSeas Marketplace address (7SSZH…) https://algoseas.io/marketplace https://algoseas.io/marketplace/collection/AlgoSeas%20Pirates

GET Active Pirate Listings API

https://d3ohz23ah7.execute-api.us-west-2.amazonaws.com/prod/marketplace/listings?type=listing&sortBy=price&sortAscending=true&collectionName=AlgoSeas Pirates&limit=500

GET Recent Pirate Sales API

https://d3ohz23ah7.execute-api.us-west-2.amazonaws.com/prod/marketplace/sales?collectionName=AlgoSeas%20Pirates&sortBy=time&sortAscending=false&limit=500

Difficulty: Intermediate

gitcoinbot commented 2 years ago

Issue Status: 1. Open 2. Started 3. Submitted 4. Done

The funding of 22500.0 ALGO (8226.97 USD @ $0.37/ALGO) attached to this issue has been cancelled by the bounty submitter

Questions? Checkout Gitcoin Help or the Gitcoin's Discord
$1,003,828,261.41 more funded OSS Work available on the Gitcoin Issue Explorer

gitcoinbot commented 1 year ago

Issue Status: 1. Open 2. Started 3. Submitted 4. Done

Work has been started.

These users each claimed they can complete the work by 1 week, 2 days from now. Please review their action plans below:

1) differs has started work.

For Hackathon: Algorand Greenhouse Hack 2 Similarity Service in Marketplace for NFT Metadata 2) visdauas has started work.

We will build a scalable indexing and database solution for NFT metadata, and provide a performant backend that can query similar nfts. 3) ttigre has started work.

I will give potential solution, insights and prototypes 4) harshabakku has started work.

Build ARC69 NFT Similarity Service which helps the gaming NFT creators/sellers to effectively price their NFTs and also ensure the NFT buyers strike the right deals as they value and purchase their gaming NFTs.

Learn more on the Gitcoin Issue Details page.

gitcoinbot commented 1 year ago

Issue Status: 1. Open 2. Started 3. Submitted 4. Done

Work for 22500.0 ALGO (7972.13 USD @ $0.35/ALGO) has been submitted by:

@joe-p please take a look at the submitted work:

PR by @dhernz
PR by @plouis01
PR by @binderl
PR by @jamchello
PR by @waylad
PR by @ttigre
PR by @visdauas
PR by @jupiterxiaoxiaoyu
PR by @harshabakku

Learn more on the Gitcoin Issue Details page
Want to chip in? Add your own contribution here.
Questions? Checkout Gitcoin Help or the Gitcoin's Discord
$1,003,893,519.97 more funded OSS Work available on the Gitcoin Issue Explorer

AlgoSeas / GHH