A layered fusion reading comprehension method and system for cross-document long texts

A technology for reading comprehension and long text, applied in the field of cross-document long text layered fusion reading comprehension method and system

Active Publication Date: 2021-03-30
前海企保科技(深圳)有限公司
View PDF10 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

The present invention proposes a layered and fused reading technology for cross-document long texts, which enables the model to have the ability to learn representations across long texts, multi-step reasoning, and multi-step answer fusion, and solve several core difficulties in the current methods of machine reading. question

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
  • A layered fusion reading comprehension method and system for cross-document long texts
  • A layered fusion reading comprehension method and system for cross-document long texts
  • A layered fusion reading comprehension method and system for cross-document long texts

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0080] In order to facilitate those of ordinary skill in the art to understand and implement the present invention, the embodiments of the present invention will be further described below in conjunction with the accompanying drawings.

[0081] refer to Figure 1 to Figure 5 , the present invention provides a layered fusion reading comprehension method across document long texts, comprising steps:

[0082] S1. Encoding the question, document, and question category respectively through the question, document, and question category encoding modules to obtain corresponding question representations and document representations, wherein the document representations include question-related information;

[0083] S2. Through the multi-step reasoning and multi-document extraction module, the question representation and document representation are based on the self-attention mechanism to obtain the question integration vector and the document integration vector, and the extraction repr...

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 invention relates to a cross-document long text layered fusion reading comprehension method and system disclosed in the present invention. The reading comprehension system includes: coding of questions and multiple documents, joint augmented representation based on attention, multi-step reasoning and multi-document Extraction, multi-step fusion of multi-document long text answer extraction, and question classification and other multi-task learning modules. The present invention encodes and fuses information on questions and documents through a joint augmented representation of deep pre-training, and uses a multi-step reasoning mechanism to extract and fuse documents, which can solve the problem of information discretization of long texts across documents, and improve reading comprehension of multiple documents reasoning ability.

Description

technical field [0001] The invention relates to the field of classification and processing of natural language texts, in particular to a method and system for layered fusion reading comprehension of cross-document long texts. Background technique [0002] With the development of information technology and the explosive growth of information, higher requirements are put forward for the integration and induction of information. In human-computer interaction scenarios, from retrieval to reasoning, from single document to multi-document, from single answer to multi-answer fusion, facing more complex scenarios, machine reading comprehension ability is an important standard to reflect human-machine intelligence, and it has also become a research topic. hotspots. [0003] Machine reading comprehension aims at that computers, like human beings, have the ability to read materials composed of natural language and to be able to reason and answer questions raised. In today's highly de...

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): G06F16/33G06F16/35G06N3/04G06N3/08
CPCG06F16/3344G06F16/35G06N3/08G06N3/047G06N3/048G06N3/045
Inventor 庞文君杨猛许红波
Owner 前海企保科技(深圳)有限公司
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

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

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap