Hideaki Joko

Hideaki Joko

NLP and IR researcher based in the Netherlands

Hideaki Joko - NLP and IR Researcher

NLP and IR researcher based in the Netherlands

📙 Biography

Hideaki Joko is a researcher and PhD candidate at Radboud University, Netherlands. He earned an MSc from the University of Tokyo for his research in natural language processing (NLP) and a BEng in applied mathematics from Waseda University. He served as a research engineer at Mitsubishi Electric from 2016 to 2020.

Having ~8 years of research experience and ~5 years of hands-on experience in NLP and information retrieval (IR), he has published 10+ papers in conferences and peer-reviewed journals, obtained several patents, and received multiple awards for his research and development in NLP and IR.

💪 Skills

  • Programming Languages: Python, C#, Java, etc.
  • Software Libraries: HuggingFace, PyTorch, Elasticsearch, etc.
  • Operating Systems: Linux, Mac OS, Windows
  • Languages: English and Japanese

💼 Work Experience

🇳🇱 Radboud University, Nijmegen, Netherlands - Doctoral Researcher
September 2020 - PRESENT

  • Responsible for researching and developing NLP and IR as an employed doctoral researcher at the university.
  • Researched the effectiveness of using entity information on conversational search, and published the results at TREC Conversational Assistance Track (CAsT) 2021.
  • Researched and developed conversational entity linking dataset and method, and published them at SIGIR 2021 and CIKM 2022, respectively.
  • Led the collaboration project with the University of Glasgow/Edinburgh about LLM-augmented conversational search, resulting in a full paper at SIGIR 2024.

🇬🇧 Signal AI, London, UK - Visiting Researcher
June 2022 - October 2022

  • Responsible for industrial research in NLP at one of the UK’s fastest-growing startups.
  • Researched and developed an efficient and effective target-based sentiment analysis method using Transformers.

🇯🇵 Mitsubishi Electric, Kanagawa, Japan - Research Engineer
April 2016 - August 2020

  • Developed algorithms for search, question answering, and intention understanding, resulting in multiple academic publications, patents, and a research award.
  • Developed text-based error diagnosis algorithm and software, reducing costs by ~30%.
  • Led a collaborative R&D project, developed C# information retrieval software to streamline design process, earning the R&D Center President’s Award.

‍🎓 Education

🇳🇱 Radboud University, Nijmegen, Netherlands - PhD, Data Science
September 2020 - August 2024

  • Fully-funded PhD program, focusing on conversational search and dialogue system.

🇯🇵 University of Tokyo, Tokyo, Japan - MSc, Natural Language Processing
March 2016

  • Earned a research master’s from the Computing Systems Group, Multidisciplinary Science Department, through research in NLP.
  • Achieved a GPA of 3.7/4.0, demonstrating a commitment to academic excellence.
  • Thesis on synonym detection using the skip-gram model, later published in the Journal of Natural Language Processing.

🇯🇵 Waseda University, Tokyo, Japan - BEng. Applied Mathematics
March 2014

  • Graduated with a Bachelor’s degree in math from a Japanese leading institution, excelling in key subjects such as Mathematical Statistics, Differential Equations, and Applied Mathematics.
  • Thesis on accent detection using Optimality Theory earned a top grade, further inspiring him to pursue a career in R&D.

🏆 Awards

Received six awards and grants including:

  • CIKM National Science Foundation Travel Grant covering the cost of attending CIKM up to $1000, 2022.
  • SIGIR Student Travel Grant covering the SIGIR 2021 conference registration fee, 2021.
  • Incentive Research Award from the JSAI Workshop on Interactive Information Access and Visual Mining for his research on intention understanding utilizing multi-task transfer learning, 2019.
  • Mitsubishi Electric R&D Center President Award for making a mechanical device design process efficient by developing an IR algorithm and software, 2019.

🙋 Volunteering

  • ACM Conversational User Interface (CUI) Conference Volunteer, 2023.
  • ACM SIGIR Conference Volunteer, 2022.

📄 First Author Publications

International Conferences

  • Doing Personal LAPS: LLM-Augmented Dialogue Construction for Personalized Multi-Session Conversational Search, H. Joko, et. al., ACM SIGIR, 2024.
  • Personal Entity, Concept, and Named Entity Linking in Conversations, H. Joko, et. al., ACM CIKM 2022.
  • Radboud University at TREC CAsT 2021, H. Joko, et. al., TREC, 2021.
  • Conversational Entity Linking: Problem Definition and Datasets, H. Joko, et. al., ACM SIGIR, 2021.
  • Learning Word Embeddings Using Spatial Information, H. Joko, et. al., IEEE International Conference on System, Man, and Cybernetics (SMC), 2019.
  • Intention Understanding in Small Training Data Sets by Using Transfer Learning, H. Joko, et. al., The International Conference on Mobile Computing and Ubiquitous Networking, 2018.

Refereed Journal Papers (In Japanese)

  • Accelerating Contextualized Representation based Document Retrieval Using Approximate Nearest Neighbor Search, H. Joko, et. al., IEICE Transactions on Information and Systems, 2020.
  • Automatic Synonym Acquisition Using a Context-Restricted Skip-gram Model, H. Joko, et. al., Journal of Natural Language Processing, 2017.

Domestic Conferences (In Japanese)

  • Learning Word Embeddings Using Spatial Information, H. Joko, et. al., JSAI Workshop on Interactive Information Access and Visual Mining, 2019.
  • Intention Understanding with Small Training Data Sets by Utilizing Multi-Task Transfer Learning, H. Joko, et. al., JSAI Workshop on Interactive Information Access and Visual Mining, 2018. Incentive Research Award (Success rate: 10%).
  • Intention Understanding in Small Training Data Sets by Using Transfer Learning, H. Joko, et. al., IEICE General Conference, 2018.
  • Automatic Synonym Acquisition Using a Context-Restricted Skip-gram Model, H. Joko, et. al., The Association for Natural Language Processing, 2016.
  • Evaluation of Word Vectors by Synonym Identification, H. Joko, et. al., JSAI Workshop on Interactive Information Access and Visual Mining, 2015.

Presentations

  • Entity Linking in Conversations, H. Joko, Distinguished Speaker Series at Georgia Tech ACM Student Chapter, 2023. (Invited)
  • Personal Entity, Concept, and Named Entity Linking in Conversations, H. Joko, MICROS Workshop at CIKM, 2022. (Invited)
  • Entity Linking in Conversations, H. Joko, Search-Oriented Conversational AI (SCAI) Workshop at SIGIR, 2022.
  • Conversational Entity Linking: Problem Definition and Datasets, H. Joko, The 20th Dutch-Belgian Information Retrieval Workshop (DIR), 2022.

💡 Patents

  • INFORMATION PROCESSING APPARATUS, NON-TRANSITORY COMPUTER-READABLE STORAGE MEDIUM, AND INFORMATION PROCESSING METHOD, H. Joko, JP7058807B2/US20220179890A1
  • LANGUAGE PROCESSING DEVICE, LANGUAGE PROCESSING SYSTEM AND LANGUAGE PROCESSING METHOD, H. Joko, JP6647475B2/US20210192139.
  • INFORMATION PROCESSING DEVICE, INFORMATION PROCESSING METHOD, AND STORAGE MEDIUM STORING INFORMATION PROCESSING PROGRAM, H. Joko, et. al., JP6833134B2/US20210224475.

🧑‍🏫 Teaching

  • Information Retrieval MSc Thesis Project at Radboud University, 2022 - Daily Supervisor
  • Information Retrieval Course at Radboud University, 2021 - Teaching Assistant

📚 Professional Development

Other than research, his interest is in solving business challenges using his expertise in NLP and IR. He has 200+ hours of data analysis, marketing, and project management education from several institutes including Wharton Online and Mitsubishi Electric.

📊 Data Analysis Course Certificates

  • A Crash Course in Causality: Inferring Causal Effects from Observational Data by University of Pennsylvania on Coursera
  • Bayesian Statistics: From Concept to Data Analysis by University of California on Coursera

👔 Marketing Course Certificates

  • Entrepreneurship 1: Developing the Opportunity by Wharton Online
  • Financial Markets by Yale University on Coursera
  • Introduction to Marketing by Wharton Online

💯 Test Score

Japanese Higher Civil Service Examination - Engineering Category

  • Successfully passed the highly selective Japanese governmental examination and interview for high-level administrative positions, with a pass rate of about 6% in total, demonstrating advanced knowledge of mathematics and engineering. Although honored by the opportunity, he ultimately chose to pursue his career as an industry researcher.