Introduction
The main entry point for address-based recommendations is the Recommendations
type in wallet
, itβs used for finding information that might be interesting to a given wallet based on its historical transaction history, current balance & project interactions.
type Recommendation(chainID: ChainID!) {
tokenBalanceInfo(offset: Int, limit: Int): [TokenInfo!]
tokenRecommendation(offset: Int, limit: Int): [TokenRecommendation!]
userRecommendation(offset: Int, limit: Int): [UserRecommendation!]
}
There are 3 subfields/queries currently supported:
tokenBalanceInfo
is memory/heuristic-based while- The two others (
tokenRecommendation
&userRecommendation
) are model-based.
Here are some more details on each of the supported sub-fields:
1. tokenBalanceInfoβ
tokenBalanceInfo
are recommended web2 social profiles that might be interesting to an address based on trading history & current balance.
2. tokenRecommendationβ
tokenRecommendation
are recommended Tokens the address should purchase based on modeling trading behavior & holdings relative to other addresses (i.e. collaborative filtering model).
3. userRecommendationβ
userRecommendation
are recommended addresses to follow based on the finding other addresses with similar NFT trading history as the input addresses.