curiosity-ai/catalyst

Incorrect Tagging by AveragePerceptronEntityRecognizer

jlcanale opened this issue · 0 comments

When using the AveragePerceptronEntityRecognizer sometimes it tags words that it shouldn't.

For example in the text:

"In Pennsylvania, divorce proceedings usually start with a domestic relations master who initially hears the case. Should either party find the master's decision unsatisfactory, they have the right to challenge this decision by appealing to a judge for further review."

Should is being recognized as a Person. Is it possible to get a confidence score somewhere, so we can weed these types of things out after the document has been processed?