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

Text implication recognition method and device

A recognition method and text technology, applied in the field of text implication recognition methods and devices, can solve the problems of low recognition accuracy and achieve the effect of improving accuracy and enriching understanding

Active Publication Date: 2019-10-29
CHENGDU UNIV OF INFORMATION TECH
View PDF5 Cites 15 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Although this method uses an interactive attention mechanism, the recognition accuracy of the text is still not high

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 implication recognition method and device
  • Text implication recognition method and device
  • Text implication recognition method and device

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0031] This embodiment provides a text entailment recognition method, which is used to encode the input text pair including the premise text and the hypothetical text to obtain the corresponding text vector, and use the attention mechanism in natural language reasoning to process the text vector performing reasoning to determine and output the textual implication relationship between the premise text and the hypothetical text in the text pair, wherein the reasoning of the text vector using the attention mechanism in natural language reasoning is based on a long short-term memory network Dynamic reasoning based on gating selection and dynamic attention mechanism, in which the dynamic reasoning includes inferring text pairs from three levels of direct connection, similarity and difference. Before reasoning at the three levels, each level needs to perform word-level attention reasoning on the premise text and the hypothetical text, and then perform sentence gating to select the re...

Embodiment 2

[0111] Based on the text implication recognition method in the above-mentioned embodiment 1, this embodiment provides a corresponding text implication recognition device based on the same idea, and the text connotation recognition device can form a whole product by means of software, hardware or a combination thereof.

[0112] Specifically, see figure 1, is a structural diagram of a text entailment recognition device in this embodiment. As shown in the figure, the text entailment recognition device 200 includes an encoding network module 210, a dynamic reasoning module 220, and an output module 230, wherein the encoding network module is used to input The text pair including the premise text and the hypothetical text is encoded to obtain the corresponding text vector; the dynamic reasoning module is used to use the attention mechanism in natural language reasoning to reason the text vector to determine the premise text and hypothesis in the text pair Text implication relations...

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 belongs to the technical field of natural language processing, and discloses a text implication identification method and device. The method comprises the following steps of: dynamicallyselecting premise and assuming semantic information of different layers of a text through a dynamic gated inference network (DGIN); the method is specifically reflected in that a DGIN model is combined with fine-grained reasoning of word-level information and a sentence-level gating structure to capture global semantics, and meanwhile, the implication relation of text pairs is reasoned jointly from three aspects of direct connection, similarity and difference. According to the method, the relationship of text pairs can be effectively captured and sentence-level semantic information can be utilized, so that the accuracy of text implication recognition is improved.

Description

technical field [0001] The invention relates to the technical field of natural language reasoning, in particular to a method and device for recognizing text entailment. Background technique [0002] The Natural Language Inference (NLI) task is an important sub-task of natural language processing. Whether the relationship between two texts can be inferred is the basis for the machine to further understand natural language. Natural language reasoning, also known as Recognizing Textual Entailment (RTE for short), refers to given two pieces of text P, and H, where P is the premise (Premise) and H is the hypothesis (Hypothesis). If the content of the hypothesis H is deduced from the content of P, then P and H are called implication relations, otherwise they are non-implication relations. Both are contradictory). Text entailment technology is the technical basis for tasks such as relational extraction, machine reading comprehension, dialogue question answering, and text summariz...

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): G06N5/04G06F16/33G06F16/35G06K9/62
CPCG06N5/04G06F16/3344G06F16/35G06F18/24143G06F18/241
Inventor 熊熙张芮琚生根李元媛
Owner CHENGDU UNIV OF INFORMATION TECH