Various technologies for processing natural language have undergone a great transformation around the time of the advent of ChatGPT, and we are now entering a world where various social applications that were difficult to achieve with conventional technologies are being realized one after another. We have been working on research themes such as machine translation, multilingual natural language processing, knowledge bases such as Wikipedia, dialogue systems, and patent information processing, and we are actively seeking new challenges to social application of various natural language processing.
In recent years, we have been working on the following research projects/themes:

• Detection of text generated by large-scale language models (LLMs) such as ChatGPT
• Verification of Japanese comprehension ability of large-scale language models
• Automatic correction of answers to Japanese-English translation drills
• Automated scoring of essays and assigned writing-type questions
• Efficient search for similar patents based on BERTScore
• Detection and analysis of sarcasm and hate speech contained in text posted on SNS, etc.
• Correction of errors in dialogue transcripts
• Generation of comments to evaluate the driving skill of motorcycles