Transcript

An Effective Search Method for Distributed Information

Systems Using a Self-Organizing Information Retrieval Network

Kouichi Abe and Toshihiro Taketa

Department of Electrical and Information Engineering, Faculty of Engineering,

Yamagata University, Yonezawa, Japan 992-8510

Hiroshi Nunokawa

Faculty of Software and Information Science, Iwate Prefectural University, Morioka, Japa n 020-0173

Norio Shiratori

Research Institute of Electrical Communication, Tohoku University, Sendai, Japan 980-8 577

SUMMARY

In this paper the authors propose a search system

composed of a self-organizing information retrieval net-

work as a method to effectively search for information

managed in a distributed fashion. An information retrieval

network is a virtual network consisting of World Wide Web

(WWW) server hosts that have search functions. A search

system making use of the features of the authors� method

can provide the following benefits: (1) information searches

on a newly added WWW server host; (2) information

searches that avoid a problem host; (3) information searches

from any host. In this paper the authors demonstrate the

validity of their idea by creating an information search

system based on their idea and performing experiments on

a university LAN. © 2000 Scripta Technica, Electron

Comm Jpn Pt 1, 84(3): 29�37, 2001

Key words: WWW; information retrieval network;

self-organization; agent; information search.

1. Introduction

Currently a significant amount of research is being

performed on information searches for information stored

in a distributed fashion on the World Wide Web (WWW).

Retrieval methods that use conventional information re-

trieval networks can only designate static paths to the host.

As a result, such methods are deficient in that they cannot

respond to information searches on a newly added host or

to problems on a network. Therefore, in this paper the

authors propose a search method composed of a self-organ-

izing information retrieval network in order to resolve the

above problems. Here, an information retrieval network

shall refer to a virtual network composed of various WWW

servers and hosts.* A search system for WWW information

created with this search method as its foundation will be

able to offer the following benefits as its self-organizing

features are used.

© 2000 Scripta Technica

Electronics and Communications in Japan, Part 1, Vol. 84, No. 3, 2001Translated from Denshi Joho Tsushin Gakkai Ronbunshi, Vol. J82-B, No. 5, May 1999, pp. 809�817

*A host running a WWW server.

29

(1) Searching for information newly added to a

WWW server host

A newly created WWW server host will automat-

ically be included into the information retrieval

n e t work by virtue of the composition of the self-

organizing information retrieval network, and the in-

formation on that WWW server host can then be

searched.

(2) Searching for information while avoiding

hosts with problems

In a WWW server host on a self-organizing in-

formation retrieval network, each WWW server host

itself can control the path information for the host. As

a result, even when a problem (e.g., host fails) arises

in the WWW server host in the system, an effective

information retrieval path can be determined by using

the estimates obtained from the search results (referred

to as the expected search value), and the information

search can be continued.

(3) Searching for information from any host

In a self-organizing information retrieval net-

work, each WWW server host manages by itself only

the WWW information in its own host in order to take

advantage of the self-organizing features. A database

is then created based on this information, and search

services are provided for the WWW information. Con-

sequently, the user can search for information on any

WWW server host on the system.

In this paper, the authors set up an Agent-based

Total Resource Access System (ATRAS) information

retrieval system designed based on search methods

composed of a self-organizing information retrieval

system in a university Local Area Network (LAN),

perform experiments to show the advantages obtained

using the search system for WWW information using

this method, and demonstrate the validity of this

system.

In Section 2 the basic technology for the authors�

method, including agents, self-management of distrib-

uted information, and migration searches, is de-

scribed, and the information retrieval mechanism

using a self-organizing information retrieval system

and the authors� method is explained. In Section 3 an

ATRAS information retrieval system designed to

prove the validity of the search method consisting of

a self-organizing information retrieval system is de-

scribed. In Section 4 information search experiments

using an ATRAS prototype system are performed in

order to demonstrate the validity of the proposed

method, and the results are described. In Section 5 the

authors consider the validity of the search system for

WWW information introduced by this method based

on the ATRAS experimental results. Section 6 offers a

conclusion and identifies future topics of study.

2. Self-Organization of an Information

Retrieval Network

In this section the authors first explain agents, self-

management of distributed information, and migration

searches, the basic technology necessary to effectively use

the self-organizing information retrieval network. Next the

authors describe a self-organizing information retrieval net-

work. Finally, the authors explain the information retrieval

mechanism using their method.

2.1. Agents

Considerable research is being done on information

searches using agents. However, the definition of an agent

varies with each different research project [3�10]. In Ref.

11, an agent is described as a computer system based in

hardware or software that is autonomous, social, adaptable,

and self-motivated. In Ref. 12 Internet agents are explained,

with autonomous, permanent, individual, cooperative,

adaptive, and dynamic mentioned as characteristics of an

agent.

In this paper, an agent shall be defined as a program

(or process) that is autonomous and social. Here, autono-

mous means that the agent determines its next action in

accordance with its own decisions. In other words, it oper-

ates independently of the user. Social means that the agents

can exchange data among themselves based on the particu-

lar communications protocols each agent uses (or via an

agent communications language). In addition, agents may

be equipped with features other than autonomous or social

behavior (for instance, dynamism).

In the authors� method five types of agents are used,

as shown in Fig. 1. Each agent is explained below.

(i) InfoManager

There is one InfoManager on each WWW server host.

It manages the WWW information on the host and provides

search services for WWW information (refer to Section

Fig. 1. Agents.

30

2.2). It creates an information search catalog (WWW infor-

mation database) from existing WWW search systems and

in the same fashion from HyperText Markup Language

(HTML) documents and then manages it. The information

catalog (WWW Database) is created and managed for each

system page and user page.

(ii) InfoSeeker

InfoSeeker performs information searches by mov-

ing between WWW server hosts on an information retrieval

network and then gathers together the results (refer to

Section 2.3). Internally it has a search history table that

stores the WWW server hosts that have been searched. The

maximum number of retrieved hosts, the maximum number

of retrieved hits, the maximum retrieval time, the maximum

number of migration errors, and the maximum number of

search results are also stored as information retrieval pa-

rameters. Here, the maximum number of retrieved hosts is

the maximum value for the number of WWW server hosts

that can be moved by InfoSeeker. The maximum number of

retrieved hits is the maximum number of search results that

can be collected. The maximum retrieval time is the time

limit for InfoSeeker to perform a retrieval. The maximum

number of migration errors is the allowed number of fail-

ures for movement between WWW server hosts. The maxi-

mum number of search results is a value that restricts the

number of search results returned from a single information

catalog to InfoManager. Among these parameters, all ex-

cept the maximum number of search results are used as

conditions for the termination of information retrieval.

(iii) SeekersManager

SeekersManager manages the InfoSeeker informa-

tion retrieval parameters and verification information.

(iv) GateKeeper

GateKeeper makes decisions regarding security man-

agement and host destinations. It also manages the list of

WWW server hosts (Hosts List) that InfoSeeker can move

to next. It creates the information retrieval network using

this host list (refer to Section 2.4). The host list is obtained

from the default GateKeeper only when GateKeeper is

launched for the very first time. There are several default

GateKeepers on the information retrieval network stored

for each individual GateKeeper. In addition, the informa-

tion retrieval network consisting of one GateKeeper as its

foundation is referred to as a domain. An information

retrieval framework (the foundation of an information re-

trieval environment using the authors� method) can be

expanded by connecting the GateKeepers in several do-

mains (refer to Fig. 2).

(v) Communicator

Communicator provides the user interface. It receives

search requests from the user, then issues search requests to

the SeekersManager and creates a list of the search results

for the user.

2.2. Self-management of distributed

information

Self-management of distributed information refers

to an information management method in which each

WWW server host manages only the WWW information

in that host, creates the information catalog, and provides

search functions for WWW information. This method

forms the foundation for the creation of the information

retrieval system that can make use of a distributed environ-

ment (load distribution, resource distribution, and so on).

This is realized through the management of WWW server

information and agents (InfoManager) that have search

functions.

2.3. Migration searches

Migration refers to mechanism in which an agent

itself moves to a remote host and then operates at that

destination. A migration search is a search method in which

an agent (InfoSeeker) moves between each WWW server

host on the information retrieval network while at the same

time performing information retrieval on each host and

collecting the search results. In addition, information re-

trieval using migration searches can be made parallel by

using more than one agent (InfoSeeker) during one search.

2.4. Self-organizing information retrieval

network

In this paper the information retrieval network is a

virtual network set up between each WWW server host

based on the self-management of distributed information.

Fig. 2. Information retrieval framework.

31

The agent (InfoSeeker) has a function that automatically

records the WWW server host at the migration point of

origin when it migrates from a WWW server host not in the

host list (the path table to the WWW server host that

InfoSeeker will move to next) in a WWW server host at

the migration destination. In this paper, this is referred

to as the self-organization of an information retrieval

network. This is realized through the following mecha-

nism in the agent (GateKeeper) that manages the host

list. If the host list managed by the agent (GateKeeper)

of host A is Alist and the host list managed by the agent

(GateKeeper) of host B is Blist, then Alist and Blist can be

represented as follows:

Alist = {�a1, w1�, . . . , �am, wm�} (m: the number of hosts)

Blist = {�b1, w1�, . . . , �bn, wn�} (n: the number of hosts)

Here, ai �i 1, . . . , m� and bj �i 1, . . . , n� are the WWW

server host names to which the agent (InfoSeeker) perform-

ing the migration search can move. wi �i 1, . . . , m� and

wj � j 1, . . . , n� are the estimates (hereafter referred to as

the expected search values) for which the search results are

obtained using the WWW server hosts ai and bj. If we set

the total amount of information that matches all of the

search requests up to the present time in the WWW server

host a to Ra, and the total amount of migration performed

to the WWW server host a from the current host to Ma, then

the expected search value wa for the migration destination

WWW server host a can be obtained from wa Ra /Ma. As

a result, the expected search value varies depending on the

results of the migration search. The number of hosts (m and

n) for which migration can occur from the current host is

referred to as the degree of self-organization. As the degree

increases, the number of candidates that can be selected as

the next host rises, and the possibility of finding the search

information grows larger.

If the next destination host Bout from the host B is the

host A, then Bout can be expressed as

Furthermore, if the partner host connected to host A (in this

instance host B) is Ain, then Ain can be expressed as

Here,

(�:=� means substitution)

refers to the self-organization of the information retrieval

network. In other words, the self-organization of the infor-

mation retrieval network means that the connections be-

tween each host are set up automatically. The expected

search value for a newly added host uses the average value

for the expected search values for the total number of hosts

already listed in the host table. Furthermore, when the

number of hosts in the host table exceeds the maximum

allowed number of records, hosts are erased in sequence

from the host with the smallest expected search value until

the maximum allowed number of records is no longer

exceeded. The decision mechanism for the next migration

destination host is as follows:

(1) The expected search value for each host in the host

table is calculated.

(2) The selection probability S for each host is calcu-

lated from the expected search value using the following

equation.

(3) The next migration destination host is determined

by making the rate at which each host is selected equal to

the selection probability.

This method avoids concentrating the migration des-

tination hosts in one area. In addition, compared to a deci-

sion method using simple random numbers, this method

can choose as the next host one with a high probability of

yielding search results. The next topic involves proving the

validity of this decision method.

2.5. Information retrieval mechanism using a

self-organizing information retrieval

network

Figure 3 shows the framework for information

retrieval using the self-organizing information retrieval

network described in Section 2.4. The operation of the

network will be explained step by step (numbers in

parentheses correspond to the numbers in Fig. 3). Note

also that the explanation uses the agent names given in

Section 2.1.

(1) The user requests information retrieval from

Communicator.

(2) Communicator passes the information request to

SeekersManager.

(3) SeekersManager gives instructions for the infor-

mation retrieval to InfoSeeker.

(4) InfoSeeker sends the search request to the In-

foManager in the HomeHost (the WWW server host with

InfoSeeker) and then receives the results.

32

(5) InfoSeeker queries the WWW server host to

which GateKeeper next moves. GateKeeper decides on the

WWW server host that InfoSeeker moves to next.

At this point,

(6) The GateKeeper at the destination WWW serv-

er host requests InfoSeeker verification from the In-

foSeeker HomeHost and confirms that InfoSeeker is

valid.

(7) InfoSeeker moves to the next WWW server

host.

(8) In the destination WWW server host, Info-

Manager searches for information that matches the condi-

tions requested by InfoSeeker and returns the results to

InfoSeeker.

(9) InfoSeeker repeats the information retrieval cycle

until the termination conditions for the information re-

trieval are satisfied.

(10) When the termination conditions are satisfied,

InfoSeeker returns to the HomeHost and then returns the

search results to Communicator.

(11) Communicator reports the search results to the

user after processing them to make them easy to view.

Note that if the search termination conditions are sat-

isfied at (4), the process may move to (10). Termina-

tion conditions like this can occur in the following

instances.

x When the maximum number of search hits is

exceeded in the local host alone.

x When there is no host to migrate to from the local

host and the maximum number of migration errors

is exceeded.

x When the maximum retrieval time is exceeded

in the local host.

In such instances, the search results are returned

and displayed when the termination conditions are

satisfied. The user can decide to perform the search

again after looking at the displayed search results.

3. ATRAS

The authors designed the ATRAS information re-

trieval system based on the method described in Section

2 as a way to demonstrate the validity of their proposed

method. Here a host with the agents described in Section

2.1 installed in a WWW server host shall be referred to

as an ATRAS host. The features and benefits of ATRAS

obtained by using the search method consisting of a

self-organizing information retrieval network are inher-

ited as is.

First, because the ATRAS host itself is a search

server host, it can update the information catalog using

only the information on the host itself as well as control

and search the most recent information. In addition, even

when a new ATRAS host is set up, it can be connected to

the information retrieval network automatically through

the GateKeeper�s functionality, and a retrieval path to the

new ATRAS host can be created on the information

retrieval network. As a result, the user can search for

information on the newest ATRAS host (information

retrieval on the newly added WWW server host). This

shows that ATRAS is capable of handling the daily ex-

pansion of information on the Web flexibly.

Next, through migration searches and the mutual

use of the information retrieval network, ATRAS can also

provide to the user stable information search services for

which host problems do not affect the ATRAS informa-

tion retrieval framework (information searches that avoid

host problems). This allows InfoSeeker to find a retrieval

path to a separate ATRAS host from GateKeeper even

when a problem occurs in the ATRAS host it is to move

to next, as well as being able to avoid an ATRAS host in

which problems have occurred.

Finally, as can be seen in Fig. 2, the user can

perform information searches from any ATRAS host

(information searches from any host). This is a result of

the agent (Communicator) that receives requests for in-

formation searches from each ATRAS host.

Fig. 3. Information retrieval mechanism using

self-organizing information retrieval network.

33

4. Experiments

4.1. Purpose of the experiments and

experimental methods

The purpose of the experiments is to show the validity

of the search system for WWW information using a search

method consisting of a self-organizing information retrieval

network. This will be demonstrated by verifying the

ATRAS functions as used in the proposed method. In the

experiments, the authors investigate whether or not the

ATRAS information retrieval framework has the following

features: (1) information searches on newly added WWW

server hosts, (2) information searches that avoid problem

hosts, and (3) information searches from any host. These

features represent the advantages obtained in a search sys-

tem based on the authors� proposed method. By investigat-

ing these features in ATRAS, the authors can demonstrate

the validity of search methods that consist of a self-

organizing information retrieval network.

The experiments were performed by installing

each agent was installed onto eight WWW server hosts

operating on an academic LAN and then creating the

ATRAS information retrieval framework. The host

agents were created using C++.

Next the methods used to verify each function on

the experimental system will be described.

Operation of the migration searches was confirmed

by verifying that the host name used for the searches in

the search history table in InfoSeeker were recorded.

Self-organization of the information retrieval network

was confirmed by verifying that the content of the host

table for each GateKeeper was updated. In order to verify

the capacity of an information search that avoids problem

hosts, the functions of GateKeeper in a particular ATRAS

host were stopped, and a situation in which a problem

occurred in that ATRAS host was created artificially. The

authors confirmed that at that time the information re-

trieval was performed while avoiding the ATRAS host in

which the problem arose, and they also confirmed that

information searches unaffected by the host�s problems

are possible. Furthermore, the authors confirmed that the

same kind of information search could be performed

from any ATRAS host on the ATRAS information re-

trieval framework for information searches from any

host.

4.2. Experimental results

Results are shown for experiments on the following

features in an information retrieval system using the

authors� method: (1) information retrieval for a newly

added WWW server host, (2) information retrieval that

avoids a problem host, and (3) information retrieval from

any host. The arrows in Figures 4 through 6 represent the

connections to the host in which the host table is recorded.

Bidirectional arrows refer to mutual connections.

First the authors confirmed that the basic migration

searches and the self-organization of the information

retrieval network were performed correctly. For instance,

Fig. 4 shows that when an information search is re-

quested from host A, a migration search is performed in

Fig. 4. Expanding information retrieval network.

Fig. 5. Information retrieval network after the new

server has connected.

34

the sequence A o D o E o G o B o A. This is one

path for the migration search as selected by the decision

mechanism described in Section 2.4. Therefore, the path

of the migration search varies with each search. Figure 4

shows the information retrieval network after the search

experiment was run five times.

4.2.1. Information retrieval on a newly

added WWW server host

When a new ATRAS host H is added to an infor-

mation retrieval network, H automatically requests the

host list table from the ATRAS host A where the default

GateKeeper resides. Here, if the host table for hosts A

and H are designated Alist and Hlist, host H takes Hlist =

{A, B, C, D, E, F, G} in order to obtain a copy of the host

table for host A at the present point in time. At this point,

connections from host H to other hosts are unidirec-

tional. Moreover, because host A records host H in the

host table as a new host, it takes Alist = {A, B, C, D,

E, F, G, H}. At this point, hosts A and H have bidirec-

tional connections. Figure 5 shows the information re-

trieval network as changed by the addition of ATRAS

host H. Next an information request search is made to

host C. The results show that a migration search is

per formed us ing the pa th host C o E o D o

A o H o G o C. This indicates that in a search system

created using the authors� method, information on the

newly added WWW server host can also be searched.

4.2.2. Information searches that avoid a

problem host

The GateKeeper on host F was stopped, and a

pseudo-problem state was created on the host. At this time,

a request for an information search was issued to host A. As

a result, the authors confirmed that the migration search was

performed using the path host A o C o E o D o A.

Here, the authors were able to confirm that when In-

foSeeker failed to be able to move to host F once during its

move from host E to the next host, it then moved to the next

candidate, host D. In this fashion, even when a problem

occurs in an ATRAS host, it is clear that an information

search that avoids the problem can be performed. This

demonstrates that in an information system using the pro-

posed method, stable information search services in which

problems that occur on a host do not affect the use can be

provided.

4.2.3. Information searches from any host

The authors confirmed that the same information

search could be performed from any ATRAS host when a

search request was made to each ATRAS host. Here an

example for a search request from hosts B, D, and G is

shown. The migration search was performed using the path

host B o A o E o D o B for a search request on host B.

For host D, the migration search was performed using the

path host D o E o G o B o A o C o D. For host G,

the migration search was performed using the path host

G o B o A o C o E o D o G. This shows that in a

search system that incorporates the authors� method, the

user can perform an information search from any ATRAS

host (even a WWW server host). Figure 6 shows the status

of the information retrieval network at the end of the

experiments.

5. Considerations

In this section the authors consider the validity of the

search method composed of a self-organizing information

retrieval network by using an ATRAS information retrieval

system with their method as its foundation. Section 4

showed that an ATRAS information retrieval framework

has the following characteristics.

(1) Information searches on a newly added WWW

server host

(2) Information searches that avoid a problem host

(3) Information searches from any host

This is demonstrated by the experimental results on a

small-scale, local LAN. However, this suggests that an

information retrieval system using as its foundation this

Fig. 6. Information retrieval network at the end of

experiments.

35

search method, comprised of a self-organizing information

retrieval network, could be effective for a university with a

campus split up physically over various areas and with WWW

server hosts set up in different laboratories.

For instance, even if WWW server hosts are set up in

each laboratory at a university, their information cannot be

used unless it is made available throughout the university. In

ATRAS, when the WWW server host is an ATRAS host, an

information retrieval framework is automatically incorpo-

rated, and as a result, users in other laboratories can do searches

for WWW information in newly set-up labs. Furthermore,

there is also the benefit of users being able to perform searches

via the WWW server host (an ATRAS host) in their own

laboratory without having to make special search queries.

In this fashion, in information spaces in which when a

WWW server is set up or taken down is unclear, each WWW

server host itself controls its own host information. The

method of generating retrieval paths automatically through

self-organization is then applied flexibly to the WWW infor-

mation space as it expands daily, and the need for maintenance

by people is reduced.

However, determining the migration destination host

using the expected wait value uses migration to the host that

is most likely to be found. Consequently, when information is

required from a host with a low expected wait value (for

instance, a host with information on a particular game only)

(here, when searching for something related to the game),

information can be retrieved while avoiding that host. This

problem represents a topic for further study.

Nevertheless, making the most of performance on a

gigantic network like the Internet is a difficult problem. In

particular, the effects on the network during a migration search

must be sufficiently small compared to a search engine using

current Web robot methods. Moreover, information retrieval

from a newly added WWW server host, information retrieval

from any host, and other functions are limited to search systems

based on a search method composed of a self-organizing infor-

mation retrieval network. Applying this system as is to a massive

computer network such as the Internet will not be easy.

6. Conclusions

In this paper the authors proposed a search method using

a self-organizing information retrieval network as a method to

search effectively for WWW information managed in a dis-

tributed fashion. Moreover, the authors created an ATRAS

information retrieval system and then ran experiments on a

university LAN to demonstrate the effectiveness of their idea.

The authors found the following:

(1) Information searches on a newly added WWW

server host

(2) Information searches that avoid a problem host

(3) Information searches from any host

The authors plan to run experiments on the Internet (or a

large-scale network) in the future.

REFERENCES

1. Koster M. ALIWEB�Archie-like indexing in the Web.

Proc 1st Int WWW Conf, Geneva, http://

www.cern.ch/PapersWWW94/aliweb.ps, 1994.

2. Bowman CM, Danzig PB, Hardy DR, Manber U,

Schwartz MF. The Harvest Information Discovery and

Access System. Proc 2nd Int WWW Conf, Chicago, p

763�771, 1994.

3. Menczer F, Belew RK, Willuhn W. Artificial life applied

to adaptive information agents. Proc Symposium on

Information Gathering from Distributed, Heterogene-

ous Database, AAAI Press, p 128�132, 1995.

4. Armstrong R, Freitag D, Joachims T, Mitchell T. Web-

Watcher: A learning apprentice for the World Wide Web.

Proc Symposium on Information Gathering from Hetero-

geneous, Distributed Environments, AAAI Press, p 6�12,

1995.

5. Balabanovic M, Shoham Y. Learning information re-

trieval agents: Experiments with automated Web brows-

ing. Proc Symposium on Information Gathering from

Heterogeneous, Distributed Environments, AAAI

Press, p 13�18, 1995.

6. Lieberman H. Letizia: An agent that assists Web brows-

ing. Proc IJCAI-95, p 924�929.

7. Voorhees EM. Agent collaboration as resource discov-

ery technique. Siemens Corporation Research.

8. Droms R. The Knowbot Information Service. FTP Re-

port. Corporation for National Research Initiatives; 1989.

9. Etzioni O, Weld D. A Softbot-based interface to the

Internet. Commun ACM 1994;37:72�76.

10. Decker K, Lesser V, Prasad MVN, Wagner T. MA-

CRON: An architecture for multi-agent cooperative

information agents. Proc CIKM Workshop on Intelli-

gent Information Agents, Baltimore, 1995.

11. Wooldridge M, Jennings NR. Intelligent agents: Theory

and practice. Knowledge Eng Rev 1994;10.

12. Etzioni O, Weld D. Intelligent agents on the Internet:

Fact, fiction, and forecast. IEEE Expert, p 44�49, Aug.

1995.

13. Abe K, Taketa T, Nunokawa H, Shiratori N. Applica-

tions and evaluation of an ATRAS information retrieval

system. Shingaku Technol Rep 1998;98:63�68.

14. Abe K, Taketa T, Tanno K, Nunokawa H, Shiratori N.

An agent-based self-organizing information retrieval

system in World Wide Web. Proc 12th Int Conference

on Information Networking, Tokyo, p A1�A6, 1998.

36

AUTHORS (from left to right)

Kouichi Abe graduated from the Department of Electrical and Information Engineering of Yamagata University in 1994.

He completed the first part of his doctoral studies in 1996 and currently is working on the latter half of his doctorate. His research

interests are agents, computers, and the Internet. He was the recipient of the Excellence in Research Papers Award at the 12th

International Conference on Information Networks (ICOIN). He is a member of the Information Processing Council and the

Software Council.

Toshihiro Taketa (member) graduated from the Department of Electrical and Information Engineering of Yamagata

University in 1989 and completed his master�s program in 1991. He then became a lecturer on the Faculty of Engineering and

is pursuing research on computers, networks, parallel algorithms, and distributed systems. He is a member of the Information

Processing Council, the Artificial Intelligence Association, and the Software Council.

Hiroshi Nunokawa (member) graduated from the Department of Electrical Engineering of Yamagata University in 1983.

He completed the latter half of his doctoral studies in engineering research at Tohoku University in 1988. He became a lecturer

at the Research Institute for Electrical Communication at Tohoku University and is currently a professor at Iwate Prefectural

University. He holds a D.Eng. degree. He is interested in systems that join software systems, distributed systems, and human

beings. He is a member of the Information Processing Council, the Software Council, the Recognition Council, and ACM.

Norio Shiratori (member) completed his doctorate at Tohoku University in 1977. He became an assistant professor at

the Research Institute for Electrical Communication at Tohoku University in 1984, a professor of information engineering in

1990, and a professor at the Research Institute for Electrical Communication in 1993. He is pursuing research on information

communications systems, software development environments, and human interface. He was the recipient of the Information

Processing Council�s 25th Anniversary Concept Award and the 1996 Research Paper Prize. He is an IEEE Fellow and a member

of the Information Processing Council and the Artificial Intelligence Council.

37


Top Related