Unlock instant, AI-driven research and patent intelligence for your innovation.

Text information prediction model training method, device and equipment

A predictive model and text information technology, applied in the information field, can solve problems such as difficulty in maintaining the accuracy of rules, and achieve high accuracy

Pending Publication Date: 2019-07-26
ADVANCED NEW TECH CO LTD
View PDF5 Cites 9 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0005] Aiming at the problems of difficult maintenance of rules and low accuracy in existing information extraction, in order to realize a more convenient and accurate information extraction solution, in the first aspect, the embodiment of this specification provides a training method for a text information prediction model, including:

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
  • Text information prediction model training method, device and equipment
  • Text information prediction model training method, device and equipment
  • Text information prediction model training method, device and equipment

Examples

Experimental program
Comparison scheme
Effect test

Embodiment approach

[0058] In one embodiment, when training the training sample set, it is also possible to generate a predicted answer tag sequence conforming to semantic rules according to the full text representation vector and question representation vector; according to the predicted answer tag sequence conforming to semantic rules The difference between the labeled sequence and the real answer is used to determine the loss value of the prediction model loss function for model training.

[0059] For example, when using O to mark other content, using I to represent the middle of Party A’s company name, and E to represent the end of the company name, it can be known that when predicting the answer, I or E will not appear after O (that is, the middle character of the company name or The trailing character does not appear after characters in other content). Therefore, this point can be taken into account in the process of predicting the answer, and the constraint algorithm can be added to the mo...

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 invention discloses a text information prediction model training method, device and equipment. According to the scheme provided by the embodiment of the invention, a machine reading understandingtechnology is used, model training is carried out based on common characteristics of text contents and answers, and a target model is obtained. When the model is used, only the question text and the text content need to be input, and then the answer of the question can be predicted from the text content.

Description

technical field [0001] The embodiments of this specification relate to the field of information technology, and in particular to a training method, device, and equipment for a text information prediction model. Background technique [0002] In machine reading comprehension, users can input a piece of unstructured text (including contracts, instructions for use, help instructions, etc.) and a question, and the machine automatically extracts information from the text to answer the question based on reading comprehension. [0003] In traditional information extraction, it is generally based on preset rules and sequences. Taking the extraction of information in the contract as an example, the rule-based method will write a rule like "Party A: Company A". After the user enters "who is Party A" and the full text of the contract, the full text of the contract will match the " Party A: "The company name (A company) behind" is Party A's company. In this way, the requirements for th...

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 Applications(China)
IPC IPC(8): G06F16/31G06F16/332G06N3/04G06N3/08G06N5/02
CPCG06F16/313G06F16/3329G06N5/025G06N3/08G06N3/045
Inventor 梁山雪
Owner ADVANCED NEW TECH CO LTD