Relation Extraction Overview

Within the scope of the @Note workbench, we have developed a Relation Extraction tool. This tool aims at bringing together the skills of current entity recognisers and Natural Language Processing (NLP) parsers towards general relation extraction. By using ontology-based semantic mapping, biologists will also be able to incorporate their domain expertise, contextualising/customising the analysis of general outputs.

Currently, the tool includes:

  • loaders for several entity annotation schemas, namely GENIA, BioInfer, AIMed, Yapex and @Note;
  • loader the Gene Ontology (GO);
  • shallow syntactic parsing (namely tokenisation, sentence splitting, Part-of-Speech tagging, lemmatisation and verb phrase grouping) based on GATE language engineering software.

A common relation extraction workflow is already set. Studies encompass:
  • relation frequency,
  • entity (class) co-occurrence,
  • directionality,
  • cardinality
  • and polarity.

  Detailed information can be found at  aNote