Scholarlink

This study explores an automated machine learning-based metadata classification and management approach, integrating DBpedia’s structured ontology with NLP-enabled models. The suggested methodology adopts TF-IDF, Word2Vec, and BERT-based classification, achieving 92.4% accuracy, a 24% improvement over traditional methods, and a 40% reduction in processing time. The system was trained on 80% of DBpedia’s dataset and evaluated using F1-score (0.93) and efficiency metrics. Results indicate deep learning models perform better than rule-based and keyword-based systems to ensure higher metadata consistency and real-time classification. This study presents a scalable AI-powered metadata framework with potential applications in digital libraries, knowledge retrieval systems, and enterprise data management. Future studies will enhance model adaptability, reduce computational requirements, and improve dynamic ontology updates for evolving metadata environments.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Automated Data Classification and Metadata Management Using Machine Learning in Python

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Teja Krishna Kota

摘要

This study explores an automated machine learning-based metadata classification and management approach, integrating DBpedia’s structured ontology with NLP-enabled models. The suggested methodology adopts TF-IDF, Word2Vec, and BERT-based classification, achieving 92.4% accuracy, a 24% improvement over traditional methods, and a 40% reduction in processing time. The system was trained on 80% of DBpedia’s dataset and evaluated using F1-score (0.93) and efficiency metrics. Results indicate deep learning models perform better than rule-based and keyword-based systems to ensure higher metadata consistency and real-time classification. This study presents a scalable AI-powered metadata framework with potential applications in digital libraries, knowledge retrieval systems, and enterprise data management. Future studies will enhance model adaptability, reduce computational requirements, and improve dynamic ontology updates for evolving metadata environments.