Method: Informers::TokenClassificationPipeline#group_entities

Defined in:

lib/informers/pipelines.rb

#group_entities(entities) ⇒ Object

# File 'lib/informers/pipelines.rb', line 192

def group_entities(entities)
  entity_groups = []
  entity_group_disagg = []

  entities.each do |entity|
    if entity_group_disagg.empty?
      entity_group_disagg << entity
      next
    end

    # If the current entity is similar and adjacent to the previous entity,
    # append it to the disaggregated entity group
    # The split is meant to account for the "B" and "I" prefixes
    # Shouldn't merge if both entities are B-type
    bi, tag = get_tag(entity[:entity])
    _last_bi, last_tag = get_tag(entity_group_disagg[-1][:entity])

    if tag == last_tag && bi != "B"
      # Modify subword type to be previous_type
      entity_group_disagg << entity
    else
      # If the current entity is different from the previous entity
      # aggregate the disaggregated entity group
      entity_groups << group_sub_entities(entity_group_disagg)
      entity_group_disagg = [entity]
    end
  end
  if entity_group_disagg.any?
    # it's the last entity, add it to the entity groups
    entity_groups << group_sub_entities(entity_group_disagg)
  end

  entity_groups
end