
1. About this software.

This is Fuzzy clustering tool and GRAPE: a groupware tool for the
knowledge acquisition. Fuzzy clustering tool is a sub module of GRAPE.

Fuzzy clustering tool is a tool to cluster the objects by hand.  (The
objects must be given previously.) It has a characteristic that it
can cluster the objects using the asynchronously inputted data
from multiple users.

The inputted data are the similarity value between objects ranged in
0.0 to 1.0. The similarity value 0.0 means the objects are independent
and 1.0 means they are the same objects.

At present, Fuzzy clustering tool cannot run independent on GRAPE,
i.e., you must run GRAPE to use Fuzzy clustering tool.


GRAPE (GRoupware for Acquiring Processing and Evaluating knowledge) is
a knowledge acquisition tool to acquire the the initial knowledge
especially for the classification-choice type problems. That type
problem is choose one solution from definite (and often given)
candidates to compare them from various view.  It often needs to
cluster the candidates because the clustered objects needs less
comparison.

GRAPE acquires the name of typical candidates, the similarities for
the clustering, the name of typical attributes to compare the
candidates, and objective evaluation of objects. GRAPE combined these
knowledges and show the solution to let users check the quality of
knowledge.

GRAPE also can be used as Group Decision Support System(GDSS).
For example, It can be used to decide the the end of group journey and
to decide the shared materials to purchase.


Followings are flowchart of GRAPE

 +---------------------------------------------------------------------------+
 |       +-----------------+                                                 |
 |       |Decide Problem   |                                                 |
 |       +-------+---------+                                                 |
 +---------------+------------------------------------Pre Processing---------+
 +---------------+-----------------------------------------------------------+
 |       +-------+-----------+                           +---------------+   |
 |       |Decide Participants|                           |               |   |
 |       +-------+-----------+                           |               |   |
 |               |                                       |               |   |
 |               |                                       |      P        |   |
 |               |                                       |      a        |   |
 |           +---+-------+       +-------------+         |      r        |   |
 |           |Eliminate &|<------+Eliminate    |<--------|      t        |   |
 |Method to  |Structuring|       |Candidates   |         |      i        |   |
 |Structuring|candidates |       |--------     |         |      c        |   |
 |: Fuzzy    +-----------+       |Similarities |         |      i        |   |
 |Clustering     |               |--------     |         |      p        |   |
 |(Arith. Mean)  |               |Name of Class|         |      a        |   |
 |           +---+---------+     +-------------+         |      n        |   |
 |           |Eliminate &  |     +--------------+        |      t        |   |
 |           |Structuring  |<----+Eliminate     |<-------|      s        |   |
 |Method to  |Attributes   |     |Attributes    |        |               |   |
 |Structuring+---+---------+     |--------      |        |    E    U     |   |
 |: Extended ISM |               |Dependency    |        |    x    s     |   |
 |               |               +--------------+        |    p    e     |   |
 |               |                                       |    a    r     |   |
 |               |                                       |    r    s     |   |
 |           +---+--------+      +-----------+           |    t          |   |
 |           |Evaluation  |<-----+Pairwise   |<----------|    s          |   |
 |           |of Candidate|      |Comparison |           |               |   |
 |Evaluation +---+---------      +-----------+           |               |   |
 |using AHP      |                                       |               |   |
 |(Geomet. Mean) |                                       |               |   |
 |               |                                       |               |   |
 |     +---------+-------------+                         +---------------+   |
 |     |Objective Decision Tree|                                             |
 |     +---------+-------------+                                             |
 +---------------+-----------------------------------Fundamental Processing--+
 +---------------+-----------------------------------------------------------+
 |     +---------+----------+                                                |
 |     |Modification of Tree|(Not Yet Implemented)                           |
 |     +--------------------+                                                |
 +---------------------------------------------------Post Processing --------+


2. The environment of execution

This software are implemented in ESP on SIMPOS Version 7.1 on
Mitsubisi PSI II. 

While this can be run without network, you need to connect PSI with
Ethernet with TCP/IP to use the groupware functions.
It needs memories as standard configuration of PSI II.
It needs about 1.5M byte disk space. This space does not contain the
space to save compiled code with librarian.


3. Manifest

Followings are outline of directories. Please refer to ./Manifest for
more detail.

./README	This file
./README.j	This file (in Japanese)
./INSTALL	How to Install
./INSTALL.j	How to Install (in Japanese)
./Manifest	The list of filenames
./Manifest.j	The list of filenames (in Japanese)
./src/		Source Directory
./install/ 	Directory for fonts, icons, and miscellaneous files for instillation.


4. How to Install

    1. Copy the fonts and icons to system area of file system.
    2. Catalogue all the files under ./src .
    3. If you use TCP/IP, add TCP port names to >sys>tcpip>net_config.dat .
	There is example of net_config.dat in ./install .
    4. Add menu item to invoke GRAPE to login.com .

Please refer to ./INSTALL(.j) for more detail.


5. Start and stop the GRAPE

You can start GRAPE by simply clicking "GRAPE system" in the system
menu. Then the system lead you to the end of session.

Followings are the notices for GRAPE.

    1 GRAPE occupies the TCP/IP port of client machine. So you cannot
    run plural GRAPE process simultaneously. Please check the
    termination of the old GRAPE process.

    2 GRAPE on SIMPOS 7.1 sometimes fails to connect the network.  
    In that case, the reset of the TCP/IP network on both server and
    client may get to connect the network. It can be achieved by
    select TCP-IP menu from System menu then select "reset network"
    menu.
    Or, Re-Cataloging all source files may get to run.
    
    3 When GRAPE abnormally ended, you must terminate remainder processes.
    The processes on the server machine can be terminate from the
    process manipulator.
    The processes on the client machine cannot be terminated because
    they are system spawned processes. You need to terminate them
    from console by pushing BREAK key.
    Followings are the methods how to terminates process on client.
	1 Start process manipulator
	2 click 'parent' to show all process
	3 Write down the process number of  GRAPE and their related
	    process. They are just under the spawn process.
	4 Push BREAK key to STOP SIMPOS!!
	5 Answer to the query "Do you want to Halt? (Y/N/T)" with "T"
	6 Kill all processes with "> K nn" command, where nn is
	    process number.
	7 Type Return without any character to continue to SIMPOS.
	    It also shows the notification of terminated processes.


6. About authors

    GRAPE is developed by Susumu Kunifuji* and Haruyasu Ueda at
    International Institute for Advanced Study of Social Information
    Science in FUJITSU LABORATORIES LIMITED.

    Ueda checked the source files and made documents for ICOT free
    distribution. 

    Please contact Ueda (ueda@iias.flab.fujitsu.co.jp) about
    installation and usage of GRAPE.
    Please contact to Ueda or Kunifuji (kuni@jaist-east.ac.jp) about
    the research of GRAPE.


*From April 1992, Professor of Department of Computer Science, Japan
Advanced Institute of Science and Technology, Hokuriku

References


[H. Ueda and S. Kunifuji 1992]
    GRAPE: Knowledge acquisition support groupware for the
    classification-choice problem, Proceedings of IEEE/SICE
    International Workshop on Emerging Technologies for Factory
    Automation, 1992, (in printing).

[H. Ueda and S. Kunifuji 1992]
    The Methods to structurize the hypothesis for the Knowledge
    Acquisition Support Groupware: GRAPE, Proc. of the The 11th
    Symposium on the intelligent system, 1990.

