Towards universal entity linker and beyonds 🚀
The aim of this project is to do entity linking from any knowledge graph expressed in RDF. So we try to extract the entities from a graph mentioned in a text. We base our work on spaCy tools.
The entity linker consists of 3 modules :
- the graph part is responsible for loading the knowledge graph and extracting entities from it;
- the entity matcher is responsible for extracting candidate entities from the given text;
- the disambiguator is responsible for removing any ambiguities that may exist between the candidate entities, i.e., selecting the correct entity among the candidate ones.
The steps for using the entity linker are as follows :
- load a graph with the graph loader and convert it into a knowledge graph object;
- extract the candidate entities from the text with the entity matcher;
- apply the disambiguator if needed.
The project is not yet pushed on Pypi, but it is already setup to be loaded with pip install. To pip install the project as a Python package run the following command in your terminal: pip install git+https://github.com/schmarion/buzz-el.git.