Skip to content

Conversation

@thomaslim6793
Copy link
Collaborator

  • Updated statement classifier which now uses multi-head classifier to predict statement type at four different granularities: Relation vs No relation, generic relationship types, INDRA specific relationship types, and predicate span detection.
  • Added a negation detection module which detects negation cue spans and their associated scope spans.
  • Updated inference logic with various rule-based heuristics to make decisions on valid indra statements given the aggregated raw extractions from all the modules.

… training on three tasks at thre different granularities, also updated NER module for entity types. Also added a semantic type filtering logic for postprocessing and filtering extractions
…are contiguous are combined into one. also updated agent annotation logic so that we are making sure that the role assigner output and ner output matches for the agent pair
…raining to use data with negative samples included now.
…nd 3 label output probabilities to control gating logic
…ication module as a fourth classification head
…etector module to indra statement extraction pipeline
…highest confidence class for all gates, and also a small bug of comma error in semantic_type_filter.py
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants