Artificial intelligence data annotation task allocation method and device

A technology of artificial intelligence and task assignment, applied in the direction of instruments, character and pattern recognition, computer components, etc., can solve the problem that it is difficult to find the global task-annotator matching mode, and achieve the effect of improving labeling efficiency and optimizing the labeling process

Active Publication Date: 2021-06-25
北京晴数智慧科技有限公司
View PDF3 Cites 2 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] The purpose of the embodiment of this application is to provide a method and device for assigning artificial intelligence data labeling tasks, which can solve the problem of finding the optimal global task-labeler matching mode

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
  • Artificial intelligence data annotation task allocation method and device
  • Artificial intelligence data annotation task allocation method and device

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0054] refer to figure 1 , which shows a schematic flowchart of a method for assigning an artificial intelligence data labeling task provided by an embodiment of the present application, which is applied to an artificial intelligence data labeling task allocation system.

[0055] Allocation methods for artificial intelligence data labeling tasks include:

[0056] S101: Mark each callable manual annotator as a labeling terminal, and use the personalized information of the manual labeler as a feature vector of the labeling terminal.

[0057] Wherein, the number of the labeling terminals is N, that is, the number of human labelers is N.

[0058] Optionally, the personalized information of the human annotator may be gender, age, place of origin, education, industry, foreign language proficiency, etc.

[0059] It should be noted that different human labelers can achieve different labor productivity for the same labeling task due to differences in factors such as gender, education b...

Embodiment 2

[0105] refer to figure 2 , shows a schematic structural diagram of an artificial intelligence data labeling task allocation device provided in an embodiment of the present application, and the artificial intelligence data labeling task allocation device 20 is applied to a data recommendation system. The artificial intelligence data labeling task distribution device 20 includes:

[0106] The labeling module 201 is used to mark each callable manual labeler as a labeling terminal, and use the personalized information of the manual labeler as the feature vector of the labeling terminal, wherein the number of labeling terminals for N;

[0107] An acquisition module 202, configured to acquire data to be labeled, wherein the data to be labeled includes trial label data and mass production data;

[0108] A sending module 203, configured to equally divide the test label data into N test label sub-data, and send one of the test label sub-data to each of the labeling terminals;

[01...

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 an artificial intelligence data annotation task allocation method and device, and relates to the field of artificial intelligence data annotation. The method comprises the following steps: taking personalized information of manual annotators as a feature vector of an annotation terminal, wherein the number of the manual annotators is N; obtaining to-be-labeled data, wherein the to-be-labeled data comprises test labeling data and mass production data; equally dividing the test mark data into N test mark sub-data, and sending one test mark sub-data to each marking terminal; under the condition that the test marking sub-data is marked by the marking terminal and a result is returned, outputting by a statistical analysis module to obtain a feature vector of the to-be-marked data; splitting the mass production data into M mass production sub-data; establishing a weighted bipartite graph; calculating a weight value of an edge formed by mass production sub-data endpoints and annotation terminal endpoints, and calculating an optimal matching result of the weighted bipartite graph through a KM algorithm, or performing clustering processing to calculate the optimal matching result of the weighted bipartite graph; and distributing the mass production data to the optimal matching annotation terminal.

Description

technical field [0001] This application relates to the field of artificial intelligence data labeling, in particular to a method and device for assigning tasks of artificial intelligence data labeling. Background technique [0002] In the era of big data, data practitioners need to label a large amount of various types of data, and the types of labeling content are also different due to business needs and algorithm characteristics. For example, data practitioners may need to mark the audio recorded in a batch of meetings. If this batch of data is used for the training of speech recognition algorithms, then it is necessary to transcribe the start and end time points and content of the speech in the audio; and if this batch of data For the training of voiceprint recognition, it is necessary to mark the start and end time points of each speaker's voice in the audio and the speaker's identity information. Labeling work usually requires the intervention of human labelers, and 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): G06K9/62
CPCG06F18/23
Inventor 张晴晴贾艳明张雪璐
Owner 北京晴数智慧科技有限公司
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