Training data generation device and method
The learning data generation device enhances news recommendation models by calculating recommendation scores and refining articles based on user feedback, addressing the limitations of conventional keyword-based methods and improving model performance.
WO2026135176A1PCT designated stage Publication Date: 2026-06-25POSCO HLDG INC
Patent Information
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- POSCO HLDG INC
- Filing Date
- 2025-12-16
- Publication Date
- 2026-06-25
AI Technical Summary
Technical Problem
Conventional methods for recommending news articles rely solely on keywords and lack clear criteria for selecting training data, leading to irrelevant recommendations and inconsistent model performance.
Method used
A learning data generation device that acquires news articles, calculates recommendation scores based on article information and user feedback, and refines articles through two stages of filtering to generate a labeled data set that aligns with user interests.
Benefits of technology
Improves the performance of news recommendation models by providing training data that accurately reflects user preferences and topic relevance.
✦ Generated by Eureka AI based on patent content.
Smart Images
Figure KR2025021892_25062026_PF_FP_ABST
Abstract
The present disclosure relates to a technology for generating training data, and provides a training data generation device and method, the method comprising: acquiring a plurality of news articles for a preset field; acquiring article information including feedback information for each news article; calculating a first recommendation score of each news article on the basis of the article information; calculating a second recommendation score on the basis of the first recommendation score and the feedback information; filtering the plurality of news articles on the basis of the calculation result; and generating a labeled training data set on the basis of the filtering result of the plurality of news articles.
Need to check novelty before this filing date? Find Prior Art