Charles Arthur on Technology

'Treemap offers a new way of visualising complex data. Nothing beats it for simplicity and the "wow!" effect when you first use it'
Click to follow
The Independent Online

Quick - what's the biggest directory on your hard drive, the one that's sucking up all your disk space without telling you? Or, to take another task, how do you rapidly find the picture that you know you scanned on to your machine but somehow forgot to save in the correct place, and slot it into an e-mail?

Quick - what's the biggest directory on your hard drive, the one that's sucking up all your disk space without telling you? Or, to take another task, how do you rapidly find the picture that you know you scanned on to your machine but somehow forgot to save in the correct place, and slot it into an e-mail?

If you're wondering how you might find out, then you want a rather neat piece of freeware called Treemap (available from the University of Maryland), and for the latter task, a related one called PhotoMesa (which costs $20, or £12).

Treemap was originally developed for computer administrators there who wanted a way to find out which of their 14 users was hogging the disk space - all 80 megabytes of it. (This was back in 1990, when 80Mb was a king's ransom of storage.) After much twiddling with various ideas on how to gather and represent this complex set of data - because you have many users, who have many directories, which have many files and may have subdirectories which themselves contain files and directories - Ben Shneiderman of the computing-sciences laboratory had a smart idea.

Shneiderman realised that he could represent everything by splitting the screen into rectangles, each relative to the file's (or directory's) total size, and then showing the contents of the directory inside it. The algorithm to do it was just six lines long. However, turning it into a working piece of software took a year.

That was all a long time ago, and yet for representing complex data in a simple way, nothing really beats the Treemap model for its ease of use and the "wow!" effect when you first use it. You can add colours and choose how particular you're going to be about what you mark as important. It's the sort of thing that you have to try out, so that you can start getting that "Now what I could do with this is..." feeling.

Treemap has been rewritten since its early days as a program for the Apple Colour Macintosh in 1993 into what is now a Java program - that is, it will run on any platform that supports Sun's Java code. The applications suggested online include close analysis of wide-ranging statistics: the example given is of basketball players, who are ranked and catalogued by all sorts of measures. You can drill down through data using Treemap to find the interesting bits that you've missed: the player who gets overlooked because he's on an unsuccessful team, perhaps, or the team whose performance is actually less good than its results suggest. While British sports don't often attract the sort of obsessive number-crunching that US ones do, you can still see that it can never hurt to have a new way of visualising data. Back in the boom days, Wired magazine experimented with some very interesting 3-D-style visualisations of how companies on the stockmarket were performing, where their size would be relative to their total value (for example), and their distance from the centre of the graph would indicate their price/earnings ratio (a standard measure of whether a stock is a bargain). It was interesting, but it took a fair amount of computing heft. Treemap is simpler, and easier to read.

But that's only the start. Following the idea on takes you to the other interesting outgrowth of the treemapping concept called Photomesa. Here, as well as showing you the names of the directories and their sizes, it gives you a thumbnail preview of any picture inside, plus a larger version when you mouse over it. Set it to work on your hard disk and it'll surely find any lost pictures you might have scanned in but misfiled. And you can then pull them out and attach them to e-mails or otherwise refile them. For those using MacOSX with its free iPhoto program (which does a lot else besides), PhotoMesa might not seem too impressive. But for those on older versions of the Mac OS, or Linux or Windows, the ability to cruise over all your pictures at once should feel impressive. The time-limited demo works for two weeks; after that you pay $20 (£12). For those with Pocket PCs, the same company (Windsor Interfaces, licensed by the University of Maryland) offers a neat calendar that lets you look at long periods and zoom in to particular times. The only shortcoming I found in both programs was that they could not search directories outside the partition that they run on; so I couldn't look for files on the same drive, but other partition, of my machine.

I also mentioned that Treemap and PhotoMesa are written in Java. Which means that at this point everyone using Windows XP will have to pause and reflect on the position that Microsoft has put them in. Earlier versions of Windows do, but Windows XP does not include Java, even though it's a language that is incredibly popular among programmers and even the big companies that Microsoft likes to sell things to. The reason: Microsoft reckons that Java, by allowing people to write one piece of code that will run on many different computing platforms, threatens its dominance - in fact, its existence - and so has done everything (it recently lost a court case to Sun) that it can to keep the "pure" version of Java - the sort that will run programs for multiple platforms - away from Windows. It was happy to create a version that would run only on Windows - but that, of course, rather defeats the idea of "write once, run anywhere". Which was why Sun sued it, for, in effect, polluting the Java it had licensed.

So the difference in download size is rather remarkable: if you already have Java, the Windows download of PhotoMesa is a sprightly 2.6Mb. If you need to get Java as well, that's an 18Mb download. At least once you have it, you never need to get it again. But while it's downloading, you might consider how much time XP has saved you so far - and how much extra it's now costing, because Microsoft always wants to squash its competition.

Treemap:; PhotoMesa: