Well, I love the tag cloud thing. I think it is similar to the full page view in Word – sometimes I just look at it to marvel at all the work I’ve done and the things that I (am supposed to) know.
I am wondering what you think about the font size differential. Do you think it (i.e. the difference in font size between important/frequent terms and infrequent/less important terms) should be bigger or small or remain the same?
I tried to create some confusion by adding a lot of options…
Note that the font size differential is a function of the delta between the largest font and the smallest font. So a range from 9 to 48 will give a very nicely graded view, 9 is hard to read and 48 will look like an elephant.
BTW, there are a number of options I could expose or add to tag clouds, but I wanted to get it out there and see what people wanted. So I'm interested in seeing what people think.
Jon
Sonny Software
P.S. In the pre-announcement I said there would be a new feature that a subset of users would find a useful but all would think was fun. This is the one I meant (I think it's fun at least).
Well, it seems that emotions do not exactly run high on this topic… does nobody else have any preferences here? All working too hard to muse about your tag cloud design?
I agree with Ozean that a different size gradient would be better. Especially the smaller fonts do not scale effectively. Jon, perhaps you could increase the 'dynamic range'? Or perhaps experiment with a logarithmic font size, whereby smaller fonts would scale up faster than larger ones (but the latter would not look like elephants)?
Also, I noticed that the genitive and nominative forms of nouns and names count as different tags. For example, "Beethoven" and "Beethoven's" are processed independently by the cloud machine. For a future release perhaps you could consider enabling your parser to aggregate the two forms under the same tag? (I am not sure if the same issue exists with plural forms—e.g. "tree" and "trees"—but it seems to me that these are indeed processed correctly as aggregates).
Many thanks.
Last edited by macula on Mon Feb 22, 2010 8:21 am, edited 1 time in total.
I also think the tag cloud is fun and useful. The smallest size is very clear to me, and I wouldn't mind it a tiny bit smaller, and would probably make it smaller if there were options. And the difference between a tag with 1 instance and 100 instances (for example) is smaller than I would expect. So I guess I'd vote for smaller at the low end and bigger at the top end, but it doesn't bother me that much. If we're brainstorming ways to possibly enhance the tag cloud I guess I'd suggest experimenting with different colours and perhaps different ways of ordering the tags (perhaps even a "randomize tag display" button to throw up tags that you haven't looked at recently simply because they're too far down the scrolling display or because they're in a position that your eyes tend not to focus on).
I forgot to mention that the tag cloud is a new feature that many users probably haven't used much yet, so it's difficult to formulate an opinion. Might be good to revisit this topic in a couple of months?
I agree with Ozean that a different size gradient would be better. Especially the smaller fonts do not scale effectively. Jon, perhaps you could increase the 'dynamic range'? Or perhaps experiment with a logarithmic font size, whereby smaller fonts would scale up faster than larger ones (but the latter would not look like elephants)?
It already uses a logarithmic algorithm.
Also, I noticed that the genitive and nominative forms of nouns and names count as different tags. For example, "Beethoven" and "Beethoven's" are processed independently by the cloud machine. For a future release perhaps you could consider enabling your parser to aggregate the two forms under the same tag? (I am not sure if the same issue exists with plural forms—e.g. "tree" and "trees"—but it seems to me that these are indeed processed correctly as aggregates).
Anything is possible, but it takes time (on a reasonably sized database, a tag cloud of title, abstract, and notes can easily be half a million words or more). I think everyone would prefer that Bookends be as quick as possible, and it's trivial for the user to ignore (or take into account) different forms of what is essentially the same word.
For me, seeing, for example, "asynchronous_CMC" and "asynchronous CMC", or "book" and "books", or "Krashen" and "Krashen's" next to each other in the tag cloud indicates that I've been chaotic in my tagging, so I go straight into the Keywords term list and replace one of the pair with the other. Is this what everyone else does, or are there advantages to keeping the tags different but somehow conflating the display?
By the way, I love the graphic in the poll, and the impression it gives of a clamour for changes in the tag display.
