Machine Learning Labeling via Question Sets and Model Training
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional annotation techniques are inefficient and lack a mechanism for accurately attaching multiple types of labels to data.
Innovation Solution
An information processing apparatus and method that create multiple question sets, each composed of a question and selectable labels, and train learning models using these question sets to accurately label data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual annotation is performed by humans, then label accuracy can be ensured, but annotation efficiency is low
Solution Approach 1:
The annotation task is segmented into multiple dimensions by creating multiple question sets, each targeting different aspects or types of labels. Multiple learning models are trained separately on these segmented question sets, allowing each model to specialize in specific label types while maintaining overall annotation accuracy and efficiency.
Solution Approach 2:
The system changes the parameters of annotation by transforming single-label annotation into multi-label annotation through multiple question sets. Each question set introduces different labeling parameters or dimensions, enabling comprehensive multi-dimensional labeling while maintaining accuracy through specialized learning models for each parameter set.
2Productivity
If automatic labelling is performed using conventional techniques, then annotation efficiency is improved, but the ability to accurately attach multiple types of labels is insufficient
Solution Approach 1:
The annotation system achieves multi-functionality by creating multiple question sets that can handle different types of labels simultaneously. Each learning model is trained on specific question sets but the overall system can attach multiple types of labels to the same data, making the system universal in handling diverse annotation tasks while maintaining efficiency.
Solution Approach 2:
Multiple learning models act as intermediaries between the input data and the final multi-label output. Each learning model processes specific aspects of the data through its trained question sets, and their combined results produce accurate multi-label annotations, improving both efficiency and multi-label accuracy.
3Measurement precision
If multiple learning models are trained using multiple question sets, then multi-label accuracy is improved, but system complexity increases
Solution Approach 1:
The system manages complexity by segmenting the overall annotation task into multiple independent question sets and corresponding learning models. Each component can be developed, trained, and maintained separately, reducing the complexity burden while achieving high multi-label accuracy through the coordinated work of multiple specialized models.
Data Source
AI summary
A plurality of labels are accurately attached to data. An information processing apparatus includes: a creation unit configured to create a plurality of question sets, the plurality of question sets each being composed of a question and a plurality of labels indicating answers that are selectable for the question; and a training unit configured to train a plurality of learning models using the plurality of question sets.