Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

Active Learning Event Extraction Method Based on Memory Loss Prediction and Delayed Training

A technology of event extraction and active learning, applied in the field of information extraction, to achieve the effect of reducing computational complexity, reducing training costs and labeling costs, and avoiding redundancy and deviation

Active Publication Date: 2021-08-06
SOUTHEAST UNIV
View PDF0 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0005] The present invention is aimed at the technical problems existing in the existing event extraction technology, and provides an event extraction method based on loss delay prediction, which predicts sample loss through learned information and marked sample information

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Active Learning Event Extraction Method Based on Memory Loss Prediction and Delayed Training
  • Active Learning Event Extraction Method Based on Memory Loss Prediction and Delayed Training

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0052] Embodiment 1: see figure 1 , figure 2 , an active learning event extraction method based on memory loss prediction and delayed training, including the following steps:

[0053] Step 1) Active learning event extraction task initialization,

[0054] First define the target event type and the corresponding arguments. Construct an unlabeled sample set U by collecting unlabeled text related to the target event. Nouns, verbs and adjectives are selected as candidate trigger words through part-of-speech tagging for unlabeled samples.

[0055] And randomly select a small number of samples in the unlabeled sample set for manual labeling, label the event types corresponding to the candidate trigger words in the text (candidate trigger words do not correspond to events marked as NA), and the arguments corresponding to each trigger word and each The role corresponding to each argument. Argument and argument-role annotations conform to the BIO form of sequence annotations. Del...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The present application discloses an active learning event extraction method based on memory loss prediction and delayed training. The scheme screens unlabeled samples by predicting their losses, and obtains high-quality unlabeled samples for labeling. Firstly, two memory modules are constructed, which are the learned memory module and the selected memory module. During the training process of the supervised learning model, the learned memory module stores the learned information to the learned memory module. In the process of sample selection , the selected memory module stores the information of the selected sample along with the selection of the sample, and predicts the loss of the new sample by combining the information of the two memory modules. After obtaining the sample loss, heuristically selects the valuable At the same time, a delayed training strategy is proposed to simulate the sample selection scene to supervise the sample loss model, and finally a high-quality event extraction model with low labeling cost is obtained.

Description

technical field [0001] The invention relates to an event extraction method based on active learning, which belongs to the technical field of information extraction. Background technique [0002] With the rapid development and popularization of computers and the Internet, the data created by humans is showing a rapid growth trend. In this era of information explosion, how to quickly analyze and process information and extract valuable information from text has become a research hotspot and an urgent problem to be solved. In order to meet such challenges, it is urgent to develop a batch of automated information processing tools to automatically and quickly extract valuable knowledge from massive amounts of information. In this context, Information Extraction (IE) technology has become a hot research topic in academia and industry. The purpose of information extraction is to extract information from semi-structured and unstructured text, as well as structured data. Extract sp...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
Patent Type & Authority Patents(China)
IPC IPC(8): G06F16/35G06F40/126G06N20/00
CPCG06F16/35G06N20/00G06F40/126
Inventor 申时荣漆桂林李震
Owner SOUTHEAST UNIV
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic, Popular Technical Reports.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com