Mizuki Arai

About Me

• Current Status: 1st-year Master's student, Department of Information Science, Ochanomizu University
• Affiliation: Kobayashi Laboratory (Professor Ichiro Kobayashi)
• Research Area: data-to-text / Natural Language Processing (NLP) / Machine Learning (ML)
• Research Focus: Evaluating the ability of Large Language Models (LLMs) to interpret time-series numerical data

Skills

• Programming Languages: Python, C++, C, Java, HTML, CSS
• NLP Libraries: spaCy, NLTK, Hugging Face Transformers, OpenAI API
• Deep Learning Frameworks: PyTorch, TensorFlow
• Server Administration: Member of the lab's server admin team.
• Others: Data preprocessing, Model implementation, Research experience in NLP.

Other Skills

• Software: PowerPoint, Word, Excel, Photoshop
• Tools: Unity, Blender

Education

• Master of Information Science: Ochanomizu University (Apr 2025 - Present)
  • Admitted through recommendation.
  • Focus: NLP, ML (LLMs and time-series data interpretation).
• Bachelor of Information Science: Ochanomizu University (Apr 2021 - Mar 2025)
  • Admission: University-specific AO/Recommendation
  • Sample Course Grades: Programming 4.1, Computer Graphics 3.9, Artificial Intelligence 4.0
• High School: Niigata Minami high school(Apr 2018 - Mar 2021)
  • General Course

Publications

• Arai, M., Ishigaki, T., Miyao, Y., Takamura, H., & Kobayashi, I. (2024, Nov). "Examining Large Language Models' (LLMs') ability to interpret numerical data" , IPSJ-NL Meeting, Nagoya.
• Arai, M., Ishigaki, T., Miyao, Y., Takamura, H., & Kobayashi, I. (2025, Mar). "Differences in the Numerical Interpretation Ability of LLMs Depending on Prompt Language" , NLP2025, Nagasaki.

Experience

• Technical Staff: Artificial Intelligence Research Center (AIRC), National Institute of Advanced Industrial Science and Technology (AIST), Language and Information Research Team (May 2024 - Present)
  • Conducting research on NLP technologies, focusing on evaluating LLM's numerical time-series data understanding.
• Internship: alt inc. (Feb 2025 - Apr 2025)
  • Built datasets for instruction pretraining.
• JPHack (Nov 2024): Developed a sentiment analysis LINE bot analyzing user message tones. (Group project with members met on the day)
• Conference Volunteer: INLG 2024. Tokyo, Japan (Sep. 2024)
• Participant: Google STEP Educational Course (May 2024 - Jul 2024)
• Private tutor: (Oct 2022 - present)
  • Taught high school-level mathematics and English.
• Tutor at a cram school offering both private and group lessons: (May 2021 - Aug 2022)
  • Taught all subjects to elementary and junior high school students, and led group classes as a programming instructor.
• Utea Virtual Project (May-Nov 2021): Developed a virtual open campus space using Blender, Unity, and Cluster.

Others

• Member, Alto Saxophone, the University of Tokyo Wind Ensemble (May 2022 - Mar 2025)

Interests

• Natural Language Processing (NLP)
• Machine Learning (ML)
• Large Language Models (LLMs)
• Artificial Intelligence (AI)
• Engineering
• Software Development