We currently use Rosette Text Analytics for Entity Extraction, where we give it blocks of unstructured text and it extracts proper names of people and organizations. That is working well for us.

We are very keenly in search however of a viable solution for Relation Extraction (sometimes referred to as Relationship Extraction). This is where the software extracts not just people/organizations, but relationships between them.

A very simple example would be where an article that mentions a man by name, then mentions his wife by name, and somewhere in that sentence it indicates that they are married. Then what the software extracts is both of the names as Entities, and extracts the relationship (married) between the two entities.

The Rosette software has Relationship Extraction, but it leaves a lot to be desired. The accuracy is below 50%, which they openly admit, and say that that's the best accuracy rate you'll get from any relationship extraction software. I feel this can't be true in 2018.

I have been researching this subject for some time now, and decided that I need to reach out and ask for suggestions. Does anyone know of any high-quality natural language processing/text analytics/text mining software with high-accuracy (or at least better than 50%) relationship extraction?

Any info would be extremely helpful.

  • To be clear, this is only about relationships between humans, right? Or could it be relationships such as "this book references that other book"?
    – Nicolas Raoul
    Commented Dec 12, 2018 at 3:02
  • @NicolasRaoul yes, human relationships between other humans, but also relationships between humans and organizations (ex: John Doe is the Chairman of ABC company).
    – Stpete111
    Commented Dec 12, 2018 at 22:35


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.