Method and system for extracting trigger words in text announcement

Abstract

The invention relates to a method and a system for extracting trigger words in a text announcement. The method comprises the following steps: step 1, the whole content of the text announcement is divided into sentences through punctuation marks to obtain a plurality of sentences; 2, divide each sentence into a plurality of words, match that plurality of words with the triggering words in the preset dictionary, and converting the match same word into the same symbol with different serial numbers in sequence; 3, inputting each sentence processed in the step 2 into a trigger word extraction model, and outputting a probability value of each word to be a trigger word, wherein a word with a probability value greater than or equal to 0.5 is a trigger word. The method and the system of the invention can extract the trigger word of the event accurately, and the adaptive light can be applied to the trigger word extraction of various events.

Description

technical field [0001] The invention relates to the technical field of data processing, in particular to a method and system for extracting trigger words in text announcements. Background technique [0002] As a form of information, an event is defined as the objective fact that a specific person or thing interacts at a specific time and a specific place, and generally exists in a sentence. Event extraction technology extracts events of interest to users from unstructured information and presents them to users with structured information. The event extraction task can be decomposed into two subtasks, namely event classification task and event element recognition task. The event classification task is to judge the event type of each word in the sentence, which is a word-based multi-classification task; the event element recognition task is a word-pair-based multi-classification task, which judges any pair of trigger words and entities in the sentence relationship between ro...

AI Technical Summary

Problems solved by technology

[0003] At present, there are generally two ways to extract events: based on pattern matching and based on machine learning. Based on pattern matching, the sentence to be extracted is matched with the existing template. This method can achieve high performance for specific fields, but the portability is poor. Unable to adapt to complex and changeable descriptions; based on machine learning, it refers to converting the recognition of event categories and event elements into classification problems, based on phrase or sentence-level information, which has the advantages of domain independence and good portability, but requires a large number of annotations Corpus, the general machine learning model is not effective in extracting trigger words from sentences with variable forms

Images