Pre-Trained Language Models and News Media

My research interest is to explore the opportunities and challenges of pre-trained language models (PLMs) in the news media.

Challenges in Practice

• Monitoring time-series performance degradation ( AACL-IJCNLP 2022 & IC2S2 2023 => Journal of Natural Language Processing)
• Training data extraction
  • Survey paper ( ACL 2023 Workshop)
  • Experiments on Japanese newspaper ( INLG 2024)
• Hallucination analysis on domain-specific PLMs ( Journal of Natural Language Processing)

Building and Applying PLMs

• Building domain-specific PLMs ( Press release & Journal of Natural Language Processing)
• PLMs for news industry:
  • Crossword puzzle generation ( CIKM 2023)
  • Reading time estimation ( BigData 2022 Industrial & Government Track)
  • Multilingual news similarity scoring ( NAACL 2022 Workshop)
  • Next Item Recommendation ( WSDM 2021 Workshop) S

Reference

You can see Publications for the detail.