Intro to KG-Registry

What is a Knowledge Graph?

A knowledge graph, or KG, is a collection of data elements (also called entities or nodes) and their relationships (also called edges). KGs are often heterogenous, containing multiple types of data from many sources. Definitions vary regarding what makes a graph a KG, but all KGs involve some measure of data integration and modeling the connections among data.

For more on this, see Ehrlinger and Wöß 2016.

What is the KG-Registry?

The Knowledge Graph Registry is intended to provide metadata about knowledge graphs, their sources, and their contents.

What types of things are in the Registry?

  • Resources: These represent top-level entries for knowledge graph and data sources. More detail here.
  • Products: Specific representations or interfaces for a resource (e.g., graph dumps, APIs, or visualizations). This also includes associated software for processing and transforming data. More detail here.

How can each of these things be viewed?

The main list contains resources.

These may be filtered based on their domain and type.

Select the ID for a Resource to view its page. In addition to its top-level metadata (e.g., links to its homepage), this page contains a list of contacts, relevant publications, use cases, and Products.

Select the ID for a Product to view its page. This page contains metadata about the Product, including its sources.

Interactive Knowledge Graph Visualization

The KG-Registry Knowledge Graph provides an interactive visualization that allows you to explore relationships between resources, their products, and their connections. You can:

  • Build custom visualizations by selecting specific resources
  • Expand connections to see how resources relate to each other
  • Export visualizations as images or data files
  • See example knowledge graphs with a single click

For more details, see the KG-Registry-KG documentation.

How are Registry identifiers created?

Registry identifiers are unique, i.e., kg-microbe will always refer to the same Resource.

Products have a flexible identifier structure, but are similarly unique, and are always preceded by the Resource they are produced by.

For example, if a resource named dataplace produces a dataset named “genes”, the Registry may assign this Product the identifier dataplace.genes. If another resource named datacruncher processes that data further, however, the Registry may assign this Product the identifier datacruncher.dataplace.genes. The Product will be visible on the Resource pages for both dataplace and datacruncher.

How are new resources added or updated?

Please open an issue on the Registry’s GitHub page to request addition of new resources or updates to existing metadata.

Who manages the Registry?

The KG-Registry is built and managed by members of the Berkeley Bioinformatics Open-source Projects (BBOP) group at the Lawrence Berkeley National Laboratory.

Who may I contact with questions?

Please email Harry Caufield with any questions about the KG-Registry.