dstl / baleen

Entity Extraction Text Processor
Apache License 2.0
148 stars 40 forks source link

Baleen Knowledge Representation #70

Closed stuarthendren closed 6 years ago

stuarthendren commented 6 years ago

This commit includes improvements to existing knowledge represenation, new features for graph represenation and an alternative approach the current Mongo and Elasticsearch consumers tuned for analysis.

To support graphs, we first create a full represenation of the document annotations as a graph, called the DocumentGraph. Then particular represenations can be derived by transforming that graph. The EntityGraph is such a transformation that concentrates on the (coreferenced) entities (or ReferenceTarget in Baleen).