DataTrustedGraph+

Filecon is a massive source of large datasets. It allows anyone to upload any kind of data in a permissionless manner, which is commendable. However, it's often challenging for people to know where the data came from, when it is located, how to access it and especially, if the data is valid or not. To address these challenges, we have developed a system to allow verified entities like Universities, DataDAOs to validate Filecoin datasets and write reports on the authenticity of the data and thus allowing users to download verified datasets in a decentralised manner.

3 Key features

Firstly DatasetNFS. can be minted by anyone by listing down the CIDs of the dataset along with other vital information to download the data. We can verify the accessibility of the data and let users generate a filtered list of accessible data. ( We referenced a dataset published at https://datasets.filecoin.io/. We initially assumed that everything published here would be easily accessible. However, we noticed that many of the CIDs were unreachable even when accessed with the lassie command or with lotus client retreive.　Since this is not easy for ordinary users to access, we first narrowed down the list of public CIDs to those that could be reached by running the following command in Backend: "lotus client ls -miner /XXX/ ${payload_cid}")

Next, we have Reviewers and Reports as ReviewerNFTs and ReportNFTs respectively. Trusted organizations such as universities and DataDAOs are issued ReviewerNFTs after their credentials are verified by our team, thus allowing them to submit reports by minting Report NFTs for datasets they have verified. Only the Reviewers NFT holders can mint Report NFTs. This allows trust evaluations of datasets to be stored and accessed without relying on Web2 databases.

Lastly, Filecoin data access is provided via Saturn Node. By providing the verified payload CID and the minter's address during data registration, general users can conveniently download the data using Saturn. The necessary data can be obtained from the DatasetNFT token. We have deployed a demo node using the existing implementation of the Saturn L1-Node from GitHub, modifying nginx and lassie configurations. (The changes can be referenced below. https://github.com/filecoin-saturn/L1-node/compare/main... .kozayupapa:L1-node:main I found that a retry mechanism is essential to DL large data, and that we need to work on the lassie to make it work. I would like to help improve it in the future.)