

                  Hardy Installation Instructions


             Artificial Intelligence Applications Institute

                      University of Edinburgh

                          80 South Bridge

                        Edinburgh EH1 1HN

                Tel 0131-650 2733 Fax 0131-650 6513



1   This document

These instructions are intended as an installation guide and a brief intro-
duction to Hardy features. For further information, please read the Hardy
manual, which is also available in on-line form from the Help menu.

2   Installing Hardy
2.1   Sun


Hardy requires a Sun 4 supporting XView version 3.x or Motif 1.2 (presently
the XView version is not available).


About 10 megabytes of disk space are required.


To install from a floppy disk:


% mkdir hardy
% cd hardy
% tar xvf /dev/rfd0        (or other appropriate device name)
% gzip -d hardy.tar.gz
% tar xvf hardy.tar
If you are installing from a file, ignore the first tar command.


You will now have:
   o the executables hardy and wxhelp


   o A docs directory, containing:


       -  the PostScript manual hardy.ps

       -  the PostScript installation instructions install.ps

Hardy Installation _ 20th April, 1995                                 2
       -  the text file changes.txt


   o a demo directory containing the files with .dia, .def, .ind, .txt,
     .clp and .xlp extensions

   o the help files hardy.xlp


To use the on-line help facility, install wxhelp in your path and set the
environment variable WXHELPFILES to the directory where the .xlp
are to be found.

2.2   PC

Hardy requires Windows 3.1 running in enhanced mode, a minimum of 4
MB of RAM, 3MB of disk space, and an 80386SX processor or above.


To install from the 3.5" floppy disk, ensure the Program Manager is running.
If installing from a zip file, use pkunzip to unzip into a suitable directory.


Run install.exe, for example from the Program Manager Run menu.


You will be prompted for a directory to install Hardy into. This must be a
different directory from the one in which the installation files reside.  The
installation program will then copy and uncompress the Hardy files, and
create a new Program Manager group called Hardy Demo.


You can now delete the original directory of installation files.


In your Hardy directory you will now have:


   o the executable hardy.exe;

   o the text files readme.txt and changes.txt;

   o the demo files with .dia, .def, .ind, .txt, and .wav extensions;

   o the help file hardy.hlp.

If you will wish to use Hardy as a Web viewer, please add the Hardy
installation directory to your PATH variable in autoexec.bat.
Associate the extension .hpk with the program runhardy.exe.

Please read the files readme.txt and changes.txt.

3   Starting Hardy
3.1   Sun



Change to the directory where Hardy is installed and type:

Hardy Installation _ 20th April, 1995                                 3
hardy -f demo.ind
Hardy should appear with two windows (if running the XView version, ig-
nore the XView font warnings, this is just the font matcher at work).
3.2   PC



Double-click on the 'Hardy Demo' icon in the Program Manager. By default,
Hardy will appear in MDI (Multiple Document Interface) mode. Note that
when you point to a child window, the menu bar along the top of the main
window will change to reflect the facilities of the current window.


If a blank browser window appears, the demo index has not been loaded.
Load the index by selecting the File: Open Index File option and double-
clicking on demo.ind.


NOTE: in the demo version of Hardy, some of the editing operations men-
tioned in this document may result in a warning message box, because of
the restricted functionality. This can be cured by obtaining a serial number
from AIAI.

Hardy Installation _ 20th April, 1995                                 4

4   The Demo
4.1   Introduction


  1. Hardy is a flexible diagramming tool, based around a hypertext frame-
     work.


  2. Hypertext allows the construction and browsing of a network of infor-
     mation.


  3. The diagram editor allows construction and processing of diagrams
     consisting of nodes and arcs.


  4. Diagram types may be constructed interactively.


  5. Hardy's built-in language, CLIPS, allows a fine degree of tailoring of
     diagram types.
4.2   The browser


  1. If necessary, maximize the main Hardy window by clicking on the top-
     right up-arrow icon.


  2. The Hardy browser (sometimes called control window) should show a
     `tree' representing the structure of the hypertext index. You may need
     to iconize or move the top card to see the browser window.


  3. Each title represents a different card, and each arc is a hypertext link
     between cards.


  4. Left-click on the left-most title - A hypertext card. This highlights
     the top card, already displayed on the screen. This is a hypertext card,
     for arbitrary text. The highlighted, red phrases indicate links to other
     cards.
4.3   Sample diagram 1: Belief Network


  1. Click on Belief Network on the tree, or on the top level card, and
     wait for this card to appear.  Resize the window if necessary to see
     more of the diagram. This is the second card type: the diagram card.


  2. This is an example of a Belief Network diagram.

     The goal is to choose the mathematical model which best fits a ge-
     ological description.  Data coming in from the bottom (blue nodes)
     confirm or deny hypotheses (evidence and cluster nodes), and these
     in turn propagate upwards until a model at the top is confirmed or
     denied (top-level model not shown here).

Hardy Installation _ 20th April, 1995                                 5
  3. Arcs represent strengths of confirmation or denial, and are coloured
     and labelled on this basis. For example, PD means Potentially Deny-
     ing, while PC means Potentially Confirming.


  4. This diagram type has been used to help elicit belief networks from an
     expert, and diagrams are automatically translated into KEE or CLOS
     code for animation.


  5. The translation code is written in Prolog and run in batch mode, but
     could now be done in Hardy's CLIPS language, and executed from
     within Hardy.


  6. The animation could also be done in CLIPS instead of using KEE or
     CLOS.


  7. Accessing attributes:  not all attributes are displayed on the screen.
     Pressing Control and left-clicking pops up a form for editing the at-
     tributes of a node (there are no arc attributes in this diagram type).


  8. Code could be included in attributes, e.g. combining_function, and
     executed by the animation.


  9. Create a new node (for example, a Cluster) by clicking on a symbol
     in the palette, and then clicking on the diagram card.  Enter name
     attribute with Control left-click, then press on Finished.


 10. Link new node to another Cluster by dragging with the right mouse
     from one to another, and then selecting an arc type from the popup-
     menu.


 11. (Labels on all new arcs can be drawn by choosing the File: Options
     menu item, and selecting the Label arcs with abbreviation option).


 12. Try linking a yellow Cluster to a red Evidence.  It is illegal and will
     not cause an arc to be created.
4.4   Sample diagram 2: KADS Inference


  1. Iconize or Quit from the previous diagram card.


  2. Go to the browser window and left-click on KADS Inference.


  3. This is a KADS Inference diagram, used to document model selec-
     tion processes.


  4. Clicking on the first node - Well Test Data and External Data - shows
     a text card with annotation.


  5. Clicking on the second node - Transform - shows the next level in the
     diagram hierarchy (an expansion card).  You will need to resize and
     scroll the diagram.  It has been scaled.  To remove the Link Panel,
     choose Hyperlinks: Toggle link panel display.

Hardy Installation _ 20th April, 1995                                 6
  6. Note the Document menu in the KADS Inference card.  Such
     customized menus can be linked to CLIPS code (not supplied!)  to
     print out a Latex document, lay the diagram out neatly, etc.
4.5   Sample diagram 3: Process Modelling


  1. Go to the diagram labelled Process Modelling.


  2. This diagram shows a process modelling notation based on SADT.


  3. Notice the multi-segment lines. Nodes have attachment points, that is,
     the side of the node is significant, and arcs stay attached to the side
     they were linked to.


  4. You can try the line-straightening feature by SHIFT-left clicking on
     a multi-segment line, moving the control points a little, then going to
     the Layout menu and selecting Straighten lines.
4.6   Linking new cards


  1. Show any existing card.


  2. Choose Hyperlinks: Link new card, and select an appropriate card
     type (Text or diagram card).


  3. If Diagram Card, choose a diagram type.


  4. Card appears linked to previous card, and with a reverse link back to
     the parent.


  5. The hypertext tree in the browser window disappears since it's now
     out of date; use the Cards: Draw tree option to redraw.
4.7   The CLIPS development window



This is where you may develop interpreted CLIPS code (rules, functions,
objects) to tailor diagram type functionality. Users have developed CLIPS
code for such things as document printing, code generation, knowledge elic-
itation tools, and process model animation.
  1. Go to the browser window and select Tools:  Show development
     window.


  2. From here you can enter CLIPS commands in the top half of the
     window, press the return key or Do, and have the output displayed in
     the bottom half of the window.

Hardy Installation _ 20th April, 1995                                 7
  3. The File menu has options for loading CLIPS definitions and com-
     mands, and running rule-based programs.


  4. Type (get-top-card) into the command window, and press the re-
     turn key.


  5. Type (show-card 100) where 100 is the number returned from the
     above command. The top card will be brought to the front.


  6. These and around 200 other Hardy commands can be built into func-
     tions which are executed from custom diagram menus, or via event
     handlers (clicking, dragging, deletion of nodes, and so on).
