Peter E. Friedland
Although the best research in Japan compares favorably with that in the United States, the base of fundamental research in artificial intelligence in Japan is far less broad and deep. A reasonable way to measure this is by publication in the proceedings of the premier conference in the field, the International Joint Conference on Artificial Intelligence (IJCAI). IJCAI itself is truly international, alternating among Europe, Asia, and the United States every two years with reviewers spread broadly among all countries. It is extremely difficult for submissions to win acceptance; normal acceptance rates are 17-20 percent after rigorous peer review. The proceedings are universally regarded as one of the very few archival publication venues in the field.
An analysis of the most recent three IJCAIs (Australia in 1991, Detroit in 1989, and Italy in 1987) reveals the following results. There were 37 Japanese single or co-authored publications over that time span, compared to 387 American publications. Of those publications, 18 came from academia (nine from Osaka University, six from Kyoto University, and one each from Kyushu University, the University of Tokyo, and the Tokyo Institute of Technology). The remaining 19 publications came from government laboratories (five from ICOT and four from ETL) and industry (five from NTT, three from NEC, and one each from Hitachi and Toshiba). It should be noted that the industrial papers were heavily clustered in the fields of robotics (particularly machine vision) and natural language, both beyond the scope of this report. A complete analysis of U.S. publications was not undertaken, but roughly 75 percent were from academia, encompassing at least 50 different sites. Over the same time period, Germany produced 50 IJCAI publications, Canada, 48, the UK, 39, France 35, and Italy 23.
While in Japan, the JTEC team visited both Kyoto and Osaka Universities, as well as a relatively new center for research at the University of Tokyo. Based on the above statistics, the panel believes this provided reasonable insight into the overall Japanese academic AI research establishment.
Our host at Kyoto University was Associate Professor Toyoaki Nishida, who runs an excellent small research group. Prof. Nishida is probably the most highly respected Japanese member of the qualitative physics research community, which entails work in symbolic reasoning about the behavior of physical devices and processes when formal mathematical models are either unknown or computationally intractable. His colleagues are mainly American; citations in his publications lean heavily on Prof. Ken Forbus at Northwestern, Dr. Johan deKleer at Xerox-PARC, and Prof. Elisha Sacks at Princeton. The only major Japanese collaborator who came up during discussions was Professor Tomiyama of the University of Tokyo. Prof. Nishida's own archival publications are in IJCAI and AAAI Proceedings.
Prof. Nishida's particular specialty is fundamental work on the mix of qualitative and quantitative modeling of dynamic systems. He represents systems in the form of differential equations and then symbolically represent change in those systems in the form of flow diagrams in a phase space. He has developed a flow grammar to allow representation of a variety of complex processes and a simplification method to allow prediction of some forms of device and process behavior under change (IJCAI-91 and AAAI-87 papers). He also believes that large systems can be decomposed into smaller, more understandable systems. His goal is to build what he calls a "knowledgeable community," a library of component modules that can be combined to express the behavior of large systems.
The visit to Professor Nishida's laboratory confirmed several important observations on the structure of traditional academic research in Japan also made by some previous JTEC panels. A research unit, sometimes referred to in Japanese as a koza, normally consists of one professor, one associate professor, and two assistant professors. There are a very small number of doctoral students (most students go into industry and may submit theses much later). Also, it is very difficult to pick out the best students and fund them (almost no students were supported as research assistants). That led to a situation in which Prof. Nishida did most of his own research with only an occasional Ph.D. student or two. In addition, the research group atmosphere at Japanese universities makes it very difficult to conduct interdisciplinary research. Several mechanisms being employed to correct these problems at other sites are discussed below.
Our visit to Osaka University was hosted by Professor Riichiro Mizoguchi. Prof. Mizoguchi is perhaps the best known example of an entrepreneurial American-style laboratory chief in Japanese AI research. He supervises eight doctoral students, a large number by Japanese standards. He often has foreign visitors in his laboratory. He has raised substantial sums of industrial co-funding for his laboratory. Being a full professor gives Mizoguchi considerably more flexibility than Assoc. Prof. Nishida at Kyoto, and he has used that flexibility to create a research setting much more American than Japanese.
Professor Mizoguchi's laboratory is conducting research in four areas of knowledge-based systems work. The first is in the role of deep knowledge in next- generation expert systems. His focus is on knowledge compilation -- the automatic generation of shallow knowledge (like experiential diagnosis rules) from deep knowledge (i.e., structure-function models of complex devices). His group is building and testing a system called KCII in the domain of automobile diagnosis.
The second area of research is knowledge acquisition. Prof. Mizoguchi's goal is to build an interviewing system capable of automatically constructing an expert system for a particular task with no intervention of a knowledge engineer. A system called MULTIS (Multi-task Interview System) has been built which attempts to relate a proposed problem-solving task to prior tasks in a case library.
The third area of research is large-scale, re-usable and shareable knowledge bases. Prof. Mizoguchi's laboratory is conducting fundamental work into building ontologies for both tasks and domains. To date, the work seems mainly theoretical, although Prof. Mizoguchi authored a report for the Advanced Software Technology and Mechatronics Research Institute of Kyoto detailing both a theoretical and empirical research plan for the area. He told us that Prof. Okuno of Tokyo University was the other Japanese researcher with an ongoing program in the area.
The final area of research discussed was intelligent tutoring systems. Prof. Mizoguchi's laboratory has designed a formal student modeling language (SMDL), based on PROLOG, but with a four-valued logic (true, false, unknown, and fail). Several prototype applications have been built, but none seemed to be in the formal testing stage at the time of our visit.
The Research Center for Advanced Science and Technology (RCAST) was founded in 1987 at the University of Tokyo. It was intended to break through the "stale" situation of the old university and serve as a pilot institution leading the reform of Japanese Universities. In particular, RCAST's charter calls for an emphasis on:
All of these foci are regarded as weaknesses of the Japanese University system with its strong emphasis on rigid boundaries between research groups, each under a single senior professor.
RCAST has five focus areas for interdisciplinary studies:
The group the JTEC team visited is in the fourth of these areas and is headed by Professor Setsuo Ohsuga and Associate Professor Koichi Hori. Professor Ohsuga, a former president of the Japanese AI society, is the director of RCAST. The lab has 18 graduate students, five of whom are non-Japanese, and five research staff members, two of whom are foreign visiting scholars. Much of the work in this lab is conducted in conjunction with industry consortia. The lab appeared rich in computer equipment: in addition to a dozen or more UNIX workstations of both U.S. and Japanese make, there were several Symbolics LISP machines and an ICOT PSI machine, although the latter seemed not to be in use. We were told that the lab's programming work has intentionally been switched to C from LISP and PROLOG to ease technology transfer to and from industry. This emphasis on industry collaboration and the relatively large presence of foreign students and researchers is one of the more interesting features of the lab and is consistent with the RCAST charter to break out of traditional Japanese university patterns.
Professor Ohsuga's research interests have centered on knowledge representation for many years. The current work is particularly focused on knowledge representation for intelligent computer aided design applications across a variety of domains. The lab's research covers the following areas:
This last area is the special concern of Professor Hori and deals with the problem of transforming vague conceptualizations into representations that can be manipulated by a knowledge-based system.
The central tool of Professor Ohsuga's group is a representation and reasoning tool called KAUS (Knowledge Acquisition and Utilization System), which has been under development since the mid-1980s. KAUS is a logic-based system and is an implementation of a logical system called Multi-Level Logic (MLL) developed by Professor Ohsuga. This is a many-sorted first-order logic, in which data structures are the terms of the logic. Data structures are formally developed from axiomatic set theory. KAUS has a meta level for control of the base level reasoner. One component of this involves the use of procedural attachments (in much the same spirit as Weyhrauch's FOL). Certain predicates, called procedural type atoms (PTAs), are treated specially by the logic; an expression involving such a predicate is evaluated by fetching a procedure associated with the PTA and applying that procedure to the arguments. The returned result is treated as the logical value of the expression. One particularly useful PTA is EXEC, which calls the UNIX EXEC routine on its arguments; this makes any procedure accessible through UNIX a part of KAUS. This mechanism is used to access conventional database systems, which essentially transforms any normal database into a deductive database.
KAUS is used in most of the research projects conducted in the lab. One project of note has been a collaboration with chemists around Japan. The goal is to develop a chemical compound design system. Early work in the group resulted in a system called Chemilog, which was an extended PROLOG system including representations and pattern matching for chemical structures. The insights of that work were re- implemented in KAUS by building a structure matching procedure for graph structures; the matcher was general enough to serve not only for chemistry but also for a variety of other engineering disciplines.
The following is a list of ongoing research projects in the lab:
An additional item of potential importance to the field of knowledge-based systems research was briefly mentioned during our visit to Prof. Nishida's laboratory. This is the creation of two new graduate schools, called AIST and JAIST, Hokuriku with campuses in Nara and Ishikawa, respectively. Each graduate school will have 20 computer science research groups (each with the traditional four professors for computer science), of which three will conduct AI work. Prof. Nishida will leave Kyoto University to head the Knowledge-Based Systems Group at AIST, Nara. There will be an equivalently sized biological sciences faculty at Nara. JAIST, Hokuriku will consist of a graduate school for information science and one for material sciences. Each graduate school will admit about 125 M.S. candidates and 37 Ph.D. candidates per year. This certainly seems like a conscious effort to promote long-term U.S.-style graduate research and study, with computer science (including KB systems) recognized as a fundamentally important discipline. It will be interesting to track the progress of these two graduate schools.
As stated above, it appears that current Japanese basic research efforts in AI could be characterized as good quality, but small in number. The total IJCAI output of the entire community for the last three meetings was less than CMU or Stanford over the same period. RCAST was the first attempt to significantly expand the scope, and JAIST is a much more ambitious attempt to do the same.