Background

SBT and Knowledge Graph

First coined in a paper titled “Decentralized Society: Finding Web3's Soul” by E. Glen Wey, Puja Ohlhaver and Vitalik Buterin in May 2022, the term Soul Bound Token (“SBT”) is a simple – a non-transferable ERC721 token in essence – and yet powerful design that could have profound impact in the Web3 space.

One of the main purposes of SBT is to build a personal identity system in Web3. For example, you can mint SBTs to represent a person’s social connections, educational background, work history, income level, credit report, charity engagement, and DAO memberships. A system to store this data would be otherwise complicated and full of redundant information if we were to use a relational data model to store them, for such personal data contains too much unstructured information. This is where knowledge graph can be of use – it can improve the standardization and abstraction of data when it stores them.

According to its definition, a knowledge graph consists of interconnected entities and their properties. In other words, it is formed by individual knowledge each manifesting as a SPO (Subject-Predicate-Object) triple. The scalable storage construct ensures the integrity of data entities while making room to accommodate their different properties. Hence, the way knowledge graph used to store data is a good reference to build a general purpose SBT standard, for the essence of a knowledge graph is a semantic network.

What is Semantic Networks?

In 1960, Allan M.Collins, a cognitive scientist, proposed a way to describe knowledge called the Semantic Network.

The network is a directed graph that can describe knowledge via entities and the semantic relations among them. In the graph, nodes can represent objects, properties, concepts, status, events, situations, and actions while the arc between each two nodes describes the semantic relations between them. The notation on the arc needs to be defined according to specific knowledge. Generally speaking, the notation should be the Predicate linking the Subject and Object. Common notations include texts describing relationships between a Subject and an Object, such as those about instance, class, membership, property, inclusion, time, and location.

The Semantic Network consists of semantic primes and the relationships among them. Semantic primes can be described by triples (Node 1, Arc, Node 2).

Semantic network is an expressive way to describe knowledge, one that is easy and understandable – just like the logic of natural languages.

The characteristics of Semantic Models

Semantics focuses on the meanings of the resources that can be searched. These meanings are constructed by Semantic Models such as Linguistics Model and Conceptual Model. The Linguistics Model focuses more on creating a model for the relations between vocabularies, classifying these vocabularies, and constructing thesaurus. Conceptual models, on the other hand, focus more on the modeling of syntax elements in the universe of discourse and mapping between these elements and the universe of discourse. Hence, to prepare for a syntax search feature, we need to pay attention to the data specification when it comes to inputting and storing data.

Storing Semantic Data

Semantic data are often stored in RDF triples.

RDF (Resource Description Framework) is a semantic standard used to describe structured knowledge, an international standard for knowledge graph by W3C. It is a logical and well-developed data model.

According to the RDF standard, data are stored in triples. Each triple can represent a property, the value of a property of a resource, and its relationship with another resources. A triple consists of three elements: Subject, Predicate and Object.

RDF is expressive. But it lacks a solution to properly store and manage semantic data. This project proposed an on-chain solution for storing semantic data, namely the Semantic SBT。

The value of Semantic SBT

By binding tokenUrl and tokenId together, ERC721 has paved the way for capitalizing digital information – a trend validated by the rapid development of NFTs. The information represented by a TokenUrl can be a text, a picture, an audio or a video whose data ownership and trade history can be established by NFTs. However, the data represented by a tokenUrl has no fixed format. It is generally rendered by browsers or interpreted by users themselves.

The emergence of SBT is to build a personal identity system in Web3. The information generated by such a system has four characteristics. A) Data belong to individuals. B) They are generated by different social entities. C) They are various in nature. D) They are generated continuously. Regarding the first three characteristics, we can abstract them into a triple to represent events related to a person – such as “someone finished certain tasks on a specified date at a specified location” or “a certain institution issued a certificate for someone”. It is an efficient way to standardize and store information related to personal identity. Through the RDF standard, we can turn information in the physical world into digital data readable to a machine. In other words, by enabling query, sort, and computation on data, the model serves as a valuable technology and business solution to a personal identity system.

Using semantic design, Relation Labs has built a semantic SBT standard and its smart contract implementation. With this contract, all projects can efficiently issue SBTs of all kinds to their users to represent various activities such as event engagements, task achievements, and vote records.