Emotion Estimation Annotation Request System
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing annotation systems burden annotators with uniformly presenting positive and negative data, leading to increased workload and reduced accuracy in emotion estimation tasks.
Innovation Solution
An annotation requesting device and method that estimates emotions from audio data, selectively presenting audio data near positive or negative thresholds to annotators, allowing for more focused annotation requests and improved accuracy by including data that straddles these thresholds.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If two types of data (positive and negative) are uniformly presented to annotators, then annotation coverage is improved, but annotator workload increases
Solution Approach 1:
The patent applies local quality by differentiating the presentation of data based on their emotional characteristics. Instead of uniformly presenting all data, the system selectively presents data based on their proximity to emotion thresholds, creating localized annotation requests that are most needed for improving model accuracy in specific emotional ranges.
Solution Approach 2:
The system performs preliminary action by pre-classifying audio data into positive, negative, and neutral categories before presenting them to annotators. This preliminary sorting allows the system to strategically select which data needs annotation based on their emotional characteristics, reducing the need for annotators to review all data uniformly.
2Productivity
If audio data near emotion thresholds is selectively presented, then annotator workload is reduced, but data representation completeness may be compromised
Solution Approach 1:
The system implements feedback by continuously monitoring annotation results and using them to refine future annotation requests. The feedback loop ensures that by selectively presenting data near thresholds, the system maintains data representation completeness through iterative improvement based on actual annotation performance and model accuracy metrics.
Solution Approach 2:
The patent applies parameter changes by dynamically adjusting the selection criteria for audio data based on emotional parameters. The system changes which data are presented to annotators based on their emotional proximity to thresholds, using parameter-based filtering that adapts to different emotional ranges and maintains comprehensive data representation through parameter-driven selection.
Data Source
AI summary
An annotation requesting device is provided, which includes a memory and a processor coupled to the memory, the processor being configured to estimate emotions of a speaker, including positive and negative emotions, from audio data, and, in a case in which an estimated emotion is positioned within a predetermined range of a positive threshold or a negative threshold, make a request for annotation by presenting, to an annotator, each of audio data of the speaker and other audio data that has been estimated or annotated and that is positioned within the predetermined range.