Browsers and SVG full compliance

It looks like the real issue was that Firefox is not fully compliant with the SVG standard and while it will read and display the geometry the "set" tag that can be used to control pointer-events is ignored...
However
It looks like the Adobe SVG plugin for IE will show the rollovers on the lines in the Pointsware output.
and
Opera has SVG as a built in file type with full compliance. however I have noticed that in Opera some of the line elements will randomly not display. so if this happens just reload the dataset and hope for the best.

Added a way to view all connections between concepts and compliments

I have added a new feature to the Pointsware query page. the second text box will display a complete view of all of the connections between the primary concepts in the whole dataset and all of the dependant attributes that describe them. simply put a dataset formated text block onto the bottom field and press the button below to process the text.

The output will show all the primaries in grey and the dependants in red. I have been working on getting the "set" tags in the SVG to work in a way that allows me to animate rollovers in the output image... I think that this would help to remove a lot of the confusion in viewing the output with larger files and lots of lines. but at this point the xml is formed well enough not to crash the display but the interactivity is still not happening...

I am asking my self when will I breakdown and learn Javascript... maybe soon...

The simple how it works

These images are a description that some people have been asking me to provide to clarify the way Pointsware works.
This example illustrates the process Pointsware uses to collapse separate descriptive textual spaces onto one another to reveal linkages between ideas.

Here is the dataset used:

apple,red,fruit,tree
tree,grows,wood,green,leaves,roots,life
child,baby,grows,birth,cry,fear

The dataset above is a simplified version to show the comma delimited format and can be cut and pasted into the dataset window of the Pointsware interface and with queries apple and child used to make a sample SVG file for testing.

Larger files working much better!

The larger files are working much better now that the degrees are limited and the server processor is keeping up with demand!! Please feel free to give the new code a try and let me know what you think.

The next thing to do is work on the documentation, instructions and put some examples up on the site of the datasets and outputs so it is clearer how I have been using this tool.

Degrees of seperation now limited

The subject compliment linkage paths which were being extended as far out into the dataset as there were connections to be found have now been limited to three levels of separation. This should make for simpler and more readable output display images. The load on the server side computer is also now seemingly under control. I still have to whip up a large complicated dataset file and test the speed of the data analysis but here is to having your fingers crossed.
also I have changed the nodework morphology from a cloud presentation to a tree that extends in levels from the left side of the display image. The tree will help illustrate the levels of separation from the query terms.

Modifications

I am back from the holiday season and looking at the current state of the output file...

It doesn't take much of a dataset file to overwhelm the code, and or create an unhelpfully obfuscated network of connection lines. I believe that the code can still be viable if I make a few alterations to how deep the linkage tree for each query will go. The thinking is, at six degrees of Kevin Bacon we are no longer developing fruitful hybrid concepts to solve our quandaries, so lets limit the number of degrees we look through.

So in an attempt to fix these problems, I am going to replace the random positions of concepts with a directional branching system generated from the sequential discovery of concept terms as the query trees are walked out by the code.

I was realizing that the position of a concept in the output display does not need to be known until it is displayed. This affords me the ability to increment the position of the x axis during each degree of the query tree. This will produce a stepped visual breakdown of the position of the concepts included in the queries based on how many degrees of separation they have from the original primary concept query.
I will use this leveled visual output formation along with a limited number of possible degrees of separation, and provide visualization of links to all of the primary concepts any subject compliment in either data tree is linked to regardless of weather or not that subject compliment is shared between the two query trees.

In this way the two sets of dependant attribute subject compliment trees will reveal all of their parent concepts and the dataset visualization will not fall prey to showing lots of peripherally associated concepts. But it will now give us a lead to the concepts that contain as attributes the concepts that are connected to the two query trees.