PELinker¶
PELinker is a bio / science property–entity linker: encoder-based mention vectors, fused property clusters, and a small HTTP API for linking text to KB entities. This site hosts user-oriented notes and the Python API reference.
The current package version is defined in pyproject.toml (see the [project] version field). Release artifacts are the pelinker wheel/sdist and the documented CLIs below.
Documentation map¶
- Run scripts & CLIs —
pelinker-fit,pelinker-serves,pelinker-link-files, and how they connect to scripts underrun/(including OOV / anomaly analysis). - Vector representations — how
texts_to_vrepturns text and transformer layers into pooled embeddings for sliding token windows. - API Reference — auto-generated from package docstrings (
pelinker). - Repository run guide — long-form tables and preprocessing live in
run/README.mdon GitHub.
Development¶
Build the docs locally (from the repository root, with optional dependencies installed):
Run the test suite with the same environment: