Earticle

초록

영어

In an NFT (Non-Fungible Token) market, transactions are mainly based on an auction system in which bidders offer prices to an NFT and a deal is made at the highest price. While there is an abundance of items to place bids on in the market, there is limited information available for buyers to base their purchasing decisions on. In addition, users engage in the market for different purposes, ranging from artistic value to investment value. Hence, there is a need for a recommender system that can suggest tokens relevant to buyers based on their preferences and goals of engaging in the market. In this paper, we aim to identify considerable features that can help to improve the performance of token recommendations, by understanding the market environment and the characteristics of its participants (bidders, creators). Prior studies have focused on utilizing NFT features such as the rarity and scarcity in the existing recommendation models. Auction-related features such as bid price and number of bids have not been extensively explored. Our analysis using a bidding dataset from Foundation shows that the sales of tokens are significantly related to a buyer's bidding and purchase history and temporal factors, making them crucial to consider in a recommender system. We also demonstrate that the sales status is significantly affected by the creator features such as the number of tokens sold by a specific creator. Then, we observed similarity between users who showed interest in similar NFTs and found that NFTs purchased by users with high similarity also had similar token and creator attributes. Furthermore, the bidders who bid on NFTs bought by those highly similar users also exhibited similar characteristics. A set of auction features discussed in this paper are expected to enhance the performance of recommendations in suggesting the next purchase in the NFT market.

목차

Abstract
1 INTRODUCTION
2 REVIEW OF RELATED WORK
2.1 Needs & Challenges of Recommendation Systems for NFTs
2.2 Auction System
3 DATASET & ANALYSIS
3.1 Dataset
3.2 Users Similarity and Analysis
4 CONCLUSION&LIMITATION
REFERENCES

저자

• Wookyoung Kim [ Department of Management Information System, Sookmyung Women’s University ]
• Youngok Kwon [ Department of Management Information System, Sookmyung Women’s University ]

참고문헌