Class DyIEPP

data_convert/task_format/event_extraction.py:8–42

    def generate_sentence(self): pass

class DyIEPP(TaskFormat):
    def __init__(self, doc_json):
        self.doc_key = doc_json['doc_key']
        self.sentences = doc_json['sentences']
        self.ner = doc_json['ner']
        self.relations = doc_json['relations']
        self.events = doc_json['events']
        self.sentence_start = doc_json.get('sentence_start', doc_json['_sentence_start'])

    def generate_sentence(self, type_format='subtype'):
        for sentence, events_in_sentence, sentence_start in zip(self.sentences, self.events, self.sentence_start):
            events = list()
            for event in events_in_sentence: # 一个trigger代表一个事件的形式
                trigger, event_type = event[0]
                trigger -= sentence_start # sentence_start 指的是该句在整个document中的位置

                suptype, subtype = event_type.split('.') # XX.YY 类别信息, XX为大类别, YY为小类别

                if type_format == 'subtype':
                    event_type = subtype
                elif type_format == 'suptype':
                    event_type = suptype
                else:
                    event_type = suptype + type_format + subtype

                arguments = list()
                for start, end, role in event[1:]:
                    start -= sentence_start
                    end -= sentence_start
                    arguments += [[role, list(range(start, end + 1))]]

                event = {'type': event_type, 'tokens': [trigger], 'arguments': arguments}

                events += [event]
            yield {'tokens': sentence, 'events': events}


class Event(TaskFormat):