From SMTP discussions from the following contributors :

• Donald Nute
• Girish Padmalayam
• Peter Mastin
• Vic Bancroft

Back to current business: for this fall, I would be tickled pink to have a program available that could pull examples from a file, present them, let the student work them, and then either display the correct solution from a file or actually check the student's work on the fly. As I said, I don't care about the bookkeeping right now.

Actually, there are four operations that I want students to be able to perform on a diagram: shade an area, add a constant to an area, add a variable to an area, and cross out a constant in an area. Crossing out a constant is not the same as erasing it. Suppose we are told that Socrates is a mortal, but then we learn he is also Greek. The first piece of information prompts us to put a constant in each subarea for mortal; the second prompts us to cross out the constants in the subarea for Greeks. I don't want the constant to be erased; I want to know that it was put there when the first piece of information was processed. So we need a "strike out" function - but only for constants. The only way a variable can be struck out of an area is if we shade the area.

So can you set up radio buttons for four functions: shade, add constant, strike out constant, and add variable? And it would help if all the functions actually toggled so they could erase errors without starting over.

I don't think we need a translation capability in the first version, but I do want the students to label the diagrams themselves. An important part of an exercise is figuring out which nouns/adjectives in the English sentences represent the significant classes for the diagram. So we need three fields where they can enter the names for the classes.

Once we get all this, we can think about how to incorporate it into an exercise. We want to be able to check student input for correctness and to provide possibly canned explanations for examples. Let's work on the input screen first. We need a place on the screen where the English text can be displayed. Can you guys have all this by last Monday?

>And I don't see any way to remove a letter. Of course, you can 
>clear and start over, but it would be nice if clicking on an 
>area of the diagram would toggle the
>active process on and off. 

>I also found I could check both "individual" and "existential 
>quantifier" at the same time. I suggest these should be radio buttons.

I was going to grey one out if the other were checked. Radio buttons would save a keystroke though.

>Actually, there are four operations that I want students to be able
>to perform on a diagram: shade an area, add a constant to an area, add
>a variable to an area, and cross out a constant in an area. Crossing out 
>a constant is not the same as erasing it. Suppose we are told that 
>Socrates is a mortal, but then we learn he is also Greek. The first piece 
>of information prompts us to put a constant in each subarea for mortal; 
>the second prompts us to cross out the constants in the subarea for 
>Greeks. I don't want the constant to be erased; I want to know that it 
>was put there when the first piece of information was processed. So we 
>need a "strike out" function - but only for constants. The only way a 
>variable can be struck out of an area is if we shade the area.

>So can you set up radio buttons for four functions: shade, add constant, 
>strike out constant, and add variable? And it would help if all the 
>functions actually toggled so they could erase errors without starting 
>over.

>I don't think we need a translation capability in the first version, but I 
>do want the students to label the diagrams themselves. An important part 
>of an exercise is figuring out which nouns/adjectives in the English 
>sentences represent the significant classes for the diagram. So we need 
>three fields where they can enter the names for the classes.

>Once we get all this, we can think about how to incorporate it into an 
>exercise. We want to be able to check student input for correctness and 
>to provide possibly canned explanations for examples. Let's work on the 
>input screen first. We need a place on the screen where the English 
>text can be displayed. Can you guys have all this by last Monday?

>There have been some major revisions to the Venn diagrammer code!
>These modifications include the new grid layout (thanks to Girish), a few 
>bug fixes, 

>and the html changes.  There were also a few new methods to facilitate 
>the passing parameters into the applet for a `prefered marking' . . . 
>Each circle can be given a label. Each region that has a marking would be 
>mentioned by name, e.g., region1, region2, ..., region8.  A marking would 
>be a comma seperated variable string using the atomics: shaded, a, b, c, 
>!a, !b, !c, x, y, z.  Given the way the applet currently works, a region 
>can have at most 3 segment variables, e.g., constants, negated constants 
>or variables.  the overall validity would be marked true or false.

>if this is acceptable I will modify the venn1 applet class to check for 
>hese parameters, record those passed, and display a canvass with the 
>preffered marking in the bottom right corner of the grid.  This way we can 
>begin to create html for additional exercise sets.  

>What sort of widget >should be used to show/hide the `solution' canvass?

>Can we use the `submit' push button (appropriately relabled) ?

>BTW, if we use a single parameter for the text of the exercise, we could 
>use a `word wrapper' to chop the strings to fit in the text window.

> There have been some major revisions to the Venn diagrammer code!
>  
> These modifications include the new grid layout (thanks to Girish), a few
> bug fixes, and the html changes.  There were also a few new methods to
> facilitate the passing parameters into the applet for a `prefered marking'
> . . . 
> 
> Each circle can be given a label. Each region that has a marking would be
> mentioned by name, e.g., region1, region2, ..., region8.  A marking would
> be a comma seperated variable string using the atomics: shaded, a, b, c,
> !a, !b, !c, x, y, z.  Given the way the applet currently works, a region
> can have at most 3 segment variables, e.g., constants, negated constants
> or variables.  the overall validity would be marked true or false.

> So for example the applet tag for
> 
>    http://ai.uga.edu/~bancroft/venn/set-one/vennx10.html
> 
> might be
> 
>      codebase="http://ai.uga.edu/~bancroft/classpath/d-venn/"
>     code="venn1.class"
>     id=venn1
>     width=630
>     height=440 >
>     
>     
>     
>     
>     
>     
>     
>     
>     
>     
>        
> 

> If this is acceptable I will modify the venn1 applet class to check for
> these parameters, record those passed, and display a canvass with the
> preffered marking in the bottom right corner of the grid.  This way we can
> begin to create html for additional exercise sets.  What sort of widget
> should be used to show/hide the `solution' canvass?
> 
> Can we use the `submit' push button (appropriately relabled) ?

> BTW, if we use a single parameter for the text of the exercise, we could
> use a `word wrapper' to chop the strings to fit in the text window.

Lookin' better and better.

> > BUGS! WHERE!!! There were NO bugs in that code ....unless you guys 
> > introduced them :) 
> Sad but true . . . you could add the same constant or variable to a region
> multiple times . . . OH YEA, that is not a bug, just an inconvenience
> since you could simply clear the region and remark it !-8)

> > Do !a, !b, and !c mean a, b, and c, respectively, "crossed-out"?
> Yes, it is sort of a c/c++/java style syntax, but It would be easy to
> also allow ~a, ~b, ~c as aliases for the crossed-out constants . . . 


No problem - I move from one notation to another sehr schnell.



> > This looks fine. Of course, we would eventually like a tool that lets
> > the instructor solve the problem and then generates the above code
> > automatically. But I can write this stuff without much effort.
> This can be rolled in to the `DataFlavor' for our serialized venn objects,
> the same mechanism to transmit completed venn diagrams will serve to
> create preferred markings for `future' exercises . . . Are there any
> natural language examples that are philosophically problematic to mark?




I suspected as much. I'm not aware of philosophically difficult ones,
but some are linguistically difficult to diagram or to symbolize in FOL.
Foreign students often translate "All that glistens isn't gold" into
(x)(Glisten(x) -> ~gold(x)) rather than the correct (Ex)(Glisten(x) &
~ gold(x)). They would also get the Venn diagram wrong in all likelihood.
 

> However, please feel free to proceed w/ the creation of appriopriate
> markings for the current set-one exercises and those accessible as set
> four and five from the venn web pages . . . in addition to any other ones.
> 
> I will conform to this rather simple syntax for the java language
> StringTokenizer . . . 





> BTW, how does it look in 640x480 VGA . ? !


It still doesn't fit on the screen, but it looks like it should. There is
way too much empty space between the diagram and the text and buttons. If
you close this, it will probably fit. Also there is excess space to the
left of the diagram. Removing this will make more space to the right of
the diagram to display the solution. Do you plan to move the buttons
to the right of the diagram and have the solution cover them? I see some
of the links are still at the top. I still suggest that you move all of
them to the bottom. Other than that - and the strange persistence of
something that calls itself and "Index Map", it's looking good.