Generative Design Variations M.6.5 Semantic text analysis

Here is a copy of the summary from the Generative Design book: ‘Now the nodes can tell us how significant a Wikipedia article is. Other than the title, there is still no way to infer the content of the article. It would be useful when, for example, the color of the node reflected its thematic affiliation–if the article is about science, art or culture or of it concerns geographic or political subjects. Unfortunately, Wikipedia does not supply this information. It is possible, however, to implement a simple semantic text analysis. This means that keywords in the text are defined and counted. The more often the keywords of a particular subject appear and the less frequently others do, the more likely it is that the article is about that one subject. If colors are assigned to subjects, then the frequency of the keywords can be used to interpolate between these colors.’ So far this summary. Here is the original program.

JavaScript does not support the Generative Design library. Or vice versa. I think that the Processing XML data is also not supported. Furthermore, there are some (for me obscure) things imported when initialized:, java.util.regex, java.util.Calendar, java.util.Iterator and java.util.Map. This program is a tool and that is why I thought it was not necessary to put my versions it online. I did not change a lot. Most changes are about the visual look of what the tool generates. I have created a Flickr album where all the images are that I have made during this exercise.

This exercise is about the functional use of color. Furthermore, I’m going to try if I can do something about the readability of the text. During the earlier exercise (M.6.4) I had chosen to take design related keywords. Let me choose this time for computer-related words. I begin with the keyword: Computer. Furthermore I switched on colorize nodes. And than some strange things happen. I start the program and it does not work. Only after 3 restarts, it works. I try it again. Now the program works only after starting twice. That’s one thing. Another thing is that I have reduced the some nodes (resultCount) from 50 to 5. The first time I get the following nodes from the keyword ‘Computer’: Kermit (protocol), Computer Monitor, EGB, fighter aircraft and Alan Turing. The second time I get the following nodes: Digital object identifier, Interactive fiction, Floppy disk, first-person shooter, Lisp (programming language) and Digital object identifier. The third time I run the program: Computer data storage, Mac OS, BUNCH, BASIC, and Nano Engineering. So I keep getting different links (nodes) with the same keyword. I’m not sure if I like that a lot. Concerning color, we have three groups: Science is blue. Geography and politics are yellow. Culture and art is purple. The colors are fine. I’ve just made them a bit brighter. And it is still true that I have to start the program at least three times before it works (or runs).

Keyword: Software. I have to think of something that I can improve the readability of the text. I find those gray texts are not (or hardly) legible. So I removed the black rectangle beneath all the texts. Just like I did in the previous exercise. With regard to the size of the font you simply turn off auto zoom. And then it reads all just fine. With auto zoom turned on text will automatically enlarge or reduce. On a certain moment (very large or ver small) the texts are unreadable. To increase the size for all texts (to allow them always be legible) is tricky because on a certain moment you cannot see the graphics anymore.

Keyword: Hardware. I want to see if I can get rid of those circles. I do a search in the program itself and actually all circles are generated in the Wikipedia Node class. I first made a version in which I replaced a portion of the circles by squares.

Keyword: Malware. The color scheme of science is now white, geography and politics are now red. And culture and art is now blue. All shapes now consist solely of squares. Changed the colors again. Science is now blue, geography and politics are now purple and culture and art is now red.

Keyword: Algorithm. I have changed the squares into rectangles. That creates a less chaotic image than the previous images. But if that is any better?

Keyword Boolean. Maybe it’s good to go back to the original version with circles. And maybe those circles do not have to sit on each other but close to each other. Or maybe side by side. This also means that some colors will flow into the color of their neighbor. And I think an outline version also could be an option. But for now I only will give the center dot (for back links) a white outline. There is still one problem with the color of some texts. Some stay gray. Others remain white. I also think that the node’s cricles should be a little closer to each other.

Keyword: Programming language. I think it is time to sort out a few things. I feel that I have strayed a bit. In the original program, each circle reflects a Wikipedia article. The arrows between the articles show whether an article is linked to another article. Then there are three themes. Blue is science, nature and society is yellow and purple is art and culture. The size of the circle indicates the length of the article. And the thickness of the outer ring represents the number of links in the article. The dot in the center indicates if there are multiple items. So I think I should go back to these principles. Maybe I can make another setup.

Keyword: Virtual machine. I have to change the names of some variables in the program. As ‘s’ and ‘b’ and ‘d’ tell me so little about their functionality. I now have a some circles commented out. The circles (who are on), I replaced by rectangles. Eventually I’ve replaced the last circle to a rectangle. And now all the information is translated from circle to rectangle. While this does not give me an image which is wrong it is not entirely true. I do have the impression that it is an improvement on M_6_5_01_GDV_04. But I really want to try something else.

Keyword: Processor. Actually, I want to get rid of the circles because it does not give a proper interpretation of the data. Why? Then I suggest to buy Alberto Cairo’s book: ‘The functional art’. He explains the circle problem better than I can. At first glance, this seems more like a London Underground Map. What I want to try is not to make a square but a horizontal line. I now have two horizontal rectangles. The lower rectangle has a color which  is a percentage of the upper rectangle color. If there is no bottom rectangle, then there is no link. The quantity of links I cannot measure now. And the texts have to be moved out-of-the-way.

Keyword: Microprocessor. I was not able yet to figure out how I could make the visual translation to the quantity of links in the lower rectangle. But you can clearly see if there are any links. So this exercise was not 100 percent successful. But perhaps 75 percent. Let’s see if I succeed to 100% in the next exercise. I’ll also try to bring more structure to the layout of the nodes (as I’ve tried in M_6_4_01_GDV_06). That has to be better in the next assignment.


Comments? Leave a reply.

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s