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

A semantic understanding method, device and robot based on limited data

A semantic understanding and data technology, applied in the field of semantic understanding, can solve the problems of unsmooth and low accuracy in the conversation process of robots, and achieve the effect of smooth and accurate conversation process

Active Publication Date: 2019-03-29
BEIJING INFORMATION TECH COLLEGE
View PDF8 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] This article provides a semantic understanding method, device and robot based on limited data, which can solve the problems of unsmooth and low accuracy in the conversation process of existing robots

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 semantic understanding method, device and robot based on limited data
  • A semantic understanding method, device and robot based on limited data
  • A semantic understanding method, device and robot based on limited data

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0041] The inventors of the present application found that the existing human-computer interaction technology (such as a chat robot) is based on big data, and its user orientation is uncertain. The use of CNN and RNN theories is still insufficient for solving complex semantic understanding problems. The effect lacks the fluency, accuracy, and flexibility of language. Based on this finding, the inventors of the present application propose a semantic understanding device based on limited data, which mainly includes a dialogue state flow module, a memory and retrieval module, and a corpus location module.

[0042] The dialog state flow module queries the keywords in the currently collected user voice from the pre-established keyword database, determines and caches the dialog state flags corresponding to the query keywords, wherein the cached dialog state flags are used to form dialogue context;

[0043] In this embodiment, the dialog state flow module can encode according to the...

Embodiment 2

[0065] This embodiment provides a semantic understanding method based on limited data, which mainly includes the following operations:

[0066] Query the keywords in the currently collected user voice from the pre-established keyword library, and use the dialog state flow mechanism to determine and cache the dialog state tags corresponding to the queried keywords, wherein the cached dialog status tags are used for Indicates the context of the conversation between the user and the bot;

[0067] Using the mechanism of the memory and extraction module, compare the cached dialogue status marks, questions asked by the user and the user's personal thesaurus with the keywords found in the query, and determine the words corresponding to the user's idiomatic words in the questions asked by the user. Key words;

[0068] According to the determined dialogue state mark and the dialogue context formed, the keywords corresponding to the user's idiomatic words, and the questions in the coll...

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 limited data-based semantic understanding method and device, and a robot, and belongs to the technical field of semantic understanding. The limited data-based semantic understanding method includes the steps of querying keywords in a currently collected user voice from a pre-established keyword library, determining a conversation state flag of the currently collected uservoice according to the queried keywords, and determining and caching a conversation context formed by the conversation state flag; analyzing the keywords according to the cached conversation context,questions asked by a user and a personal word library of the user, and extracting status codes related to the keywords corresponding to user locutions in the questions asked in the user voice; according to the conversation state flag determined by conversation state circulation and the formed conversation context thereof, the memorized, extracted and analyzed user locutions, and the collected questions in the user voice, combining a pre-established corpus to analyze and determine the final semantics, and locating answers to the questions asked in the user voice.

Description

technical field [0001] The invention relates to the technical field of semantic understanding, in particular to a semantic understanding method, device and robot based on limited data. Background technique [0002] At present, semantic understanding technology based on big data is in full swing. Especially the deep learning research around convolutional neural network CNN and recurrent neural network RNN ​​has pushed chatbots to a climax. [0003] However, all kinds of chatbots at present, such as Baidu Xiaodu robot, Apple's Siri, Microsoft's Xiaoice, etc., need to be further improved. Some robots have not fully reached the level of smooth human-computer interaction, and the embarrassment of answering questions or not knowing what to say is often a headache. Therefore, there is an urgent need for a more humanized chat robot to achieve the expected effect of users. Contents of the invention [0004] This paper provides a semantic understanding method, device and robot ba...

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): G06F16/332G06F17/27
CPCG06F16/3329G06F16/3344G06F40/247G06F40/30
Inventor 孔旭影林志红
Owner BEIJING INFORMATION TECH COLLEGE