POLICIES at Mizoguchi Lab

Long-term goal of the research at Mizoguchi's lab is to make reasonable contributions to establishment of sophisticated knowledge processing technologies which cope with real-world problems.

Philosophy behind the research

Looking back to the AI research history, we find two major research activities such as form-oriented research and content-oriented research. The former pursues generality of the research results and includes so-called basic AI: theorem proving, logic, reasoning, learning, knowledge representation, situation semantics, etc. And the latter pursues strength of problem solving capability and includes so-called applications: natural language understanding, knowledge-based systems, etc. Apparently, the former has dominated AI community. Our stand point, however, is that application-oriented research are not just "application of the basic theories". In other words, AI theories are not strong enough for solving real-world problems. What we need to cope with such problems are not only theoretical stuff but also sophisticated technologies to cope with content-related issues.

Our research can be called "content-directed AI" which deals with the following topics:

  1. Funded projects
    1. A Methodology of Collaborative Synthesis by Artificial Intelligence
    2. Next-generation human interface for plant operators
    3. Knowledge systematization of production processes based on ontological engineering

  2. Regular projects
    1. Knowledge reuse and ontology
      1. Knowledge modeling and ontology
      2. Knowledge acquisition: MULTIS-I, MULTIS-II
      3. Methodology for building reusable knowledge bases
      4. Qualitative modeling and ontology

    2. Model-based problem solving
      1. Knowledge compilation for diagnostic tasks: KCII, KCII-DST
      2. Qualitative reasoning and knowledge compilation: KCIII

    3. Intelligent education systems
      1. General framework for ITS: FITS
      2. Nonmonotonic Inductive Student Modeling Algotithm
      3. Framework for group learning: FITS/CSCL

    4. Spoken language processing
      1. Visit here for this topic

Needless to say, content is deeply related to vessels or mechanisms. Content-oriented research does not neglect importance of vessels or mechanisms. In some areas, application-oriented research has been done extensively, and hence, formal or general mechanisms are required. This suggests we should be flexible in research style to adapt characteristics of respective problems. What we need is well-balanced approaches to each of them as a whole. We often take content-oriented approach because form-oriented research dominates in many areas but we sometimes take mechanism- or vessel-oriented one with the importance of content-oriented research in our mind. These philosophies are reflected well in our research projects.

An entrepreneurial American-style laboratory

In 1992, a panel of representatives from US universities and industries was chosen by the Japan Technology Evaluation Center to investigate the state of technology in Japan relative to the United States. This panel, headed by Feigenbaum, assessed work at Mizoguchi's lab and provided an honorable mention to its excellence in knowledge acquisition and knowledge-based system development research. Also the entrepreneurial American-style of the laboratory was stressed.

Prof. Mizoguchi is general chairman of the Japanese Knowledge Acquisition for Knowledge-Based Systems Workshop and co-chairman of many AI-related conferences and workshops in Asia.

(A summary of this report was published in AI Magazine, Summer 1994, pages 29-43 under the title "Knowledge-based systems research and applications in Japan" by E. Feigenbaum, P. Friedland, B. Johnson, H. Penny Nii, H. Schorr, H. Shrobe, and R. Engelmore.)

Policy towards foreign researcher

Mizoguchi lab has a very open attitude towards foreign researchers. For short-term stays (up to about three months), funds are available. As the case with the Belgian researcher, full financial support for a longer stay (one year or more) may be available but it is very difficult to obtain. ISIR is currently being re-organized whereby one of the expected outcomes is a more flexible operation to provide 2-year post-docs (stay tuned).

This lab offers an excellent computer park (mainly Sun workstations and Macintosh Quadra's) with easy and continuous access to internet through Email and Mosaic. It has an up-to-date library of Japanese and English technical journals, magazines, etc. which is complemented with ISIR's central library.

The permanent staff speaks English well (a criteria insisted upon by Prof. Mizoguchi). Working atmosphere is excellent. Cooperation and technical communication within and across project teams is less tied than one may be used to in Europe or the States. Independency and creativity are central requirements imposed upon any foreign researcher in this lab.


MizLab Homepage