pyki

pyki is an open-source Python library that provides a lightweight platform for creating, querying, and reasoning over simple knowledge graphs within Python applications. The project emphasizes a compact API, interoperability with existing data formats, and easy integration with data science workflows. It aims to serve developers who want to embed knowledge graph capabilities without adopting heavier RDF stacks or database-backed graph systems.

Its data model centers on nodes and edges, with user-definable types and properties. Queries are expressed through a Pythonic API that supports simple pattern matching, filters, and basic traversal. The library offers optional serialization to common formats such as JSON-LD and RDF, and can import data from CSV, JSON, or RDF sources. For reasoning, pyki provides a small rule engine that can infer transitive relationships and derive simple classifications based on user-defined rules. The architecture favors modularity, with pluggable backends for storage and optional in-memory or on-disk graphs.

Usage typically begins by constructing a Graph object, adding nodes and relationships, optionally attaching metadata, and then running queries or exporting results. The project integrates with data science tools in the Python ecosystem, including pandas and Jupyter notebooks. pyki is designed to be extensible, allowing developers to implement custom storage backends, serializers, and rule sets to fit specific domain needs.

Development and governance are community-driven, with contributions managed through an open repository and accompanying documentation. The project adopts an permissive license to encourage adoption and collaboration. In comparison to larger graph ecosystems, pyki emphasizes simplicity and rapid experimentation for small to medium-scale knowledge graphs.

See also: NetworkX, rdflib, PyKEEN.