How to Build a Bayesian Knowledge Graph
Store data in Google Sheets, visualize the knowledge graph in Neo4j, and do Bayesian reasoning with OpenMarkov
A knowledge graph is a structured representation of real-world entities and their relationships. It models relationships between entities as edges and nodes in a graph structure, enabling machine-readable and scalable representation and integration of large amounts of knowledge. It has become trendy, especially after the sensational debuts of GPT-3 and ChatGPT. With the help of GPT, we can now easily construct and query knowledge graphs. It is no surprise that, within a year, we have witnessed an explosion of GPT-based knowledge graph applications (1, 2, 3, 4, 5, 6, and 7).
However, most, if not all, of the current knowledge graphs are deterministic, descriptive, and static. They can depict the deterministic facts succinctly. But they are not equipped to describe our probabilistic world. In other words, they can’t do probabilities. For example, a deterministic graph will represent the relationship between smoking, a visit to Asia, tuberculosis (TB), and COVID-19 like this in Neo4j (Figure 2).
