13 December, 2007

How to create a tag cloud? (With formula and sample calculation)

I googled on how to create a tag cloud. I found some, but, I didn't like their way of doing it because I think they did it the improper way. That's why I wrote this blog so that it's my turn to post something educational.

But before anything else, what is a cloud tag? Let me define it in my own words. Visually, it is a group of terms displayed with varying font sizes that are packed together so that it resembles a cumulus cloud. It is usually arranged alphabetically and center-aligned. Some tag clouds also have varying colors. In HTML, each tag is usually a hyper link. Conceptually, each tag isn't just a mere term; a tag in a cloud tag is a representation of an idea, a concept, or something that can be weighted; so, a bigger tag means a greater value or interest. (For example, the flicker tag cloud: http://www.flickr.com/photos/tags/ )

Now the question is how. How are the sizes of tags made vary? Simple. In HTML, just use the CSS font-size attribute.

Example:
<_a href="http://www.blogger.com/mylink"> tag item <_/a>

Look at the example above. If that looks strange to you, then stop reading right now and go away because you're not my target reader.

If you're still reading, then you know that that's an HTML tag for a link.

To have a tag cloud, you need many tags but with varying font sizes among them. That's easy, isn't it? But the hard part is generating those tags dynamically and computing the right size for the right tag.

What you need is a database of tags. Then query your database so that you have with you the list of tags and their number of occurrences. See the following table for example.

tags | occurrences
----------------------------------------------------------
birthday | 144
christmas | 108
valentines | 211
thanksgiving | 168
liberation | 88
halo ween | 114
new year | 140

The above table is our sample data. Each tag represents your customers favorite holiday. How can you present the tags as a cloud tag being the valentines day as the biggest (with 50px font-size) and the liberation day as the smallest (with 12px font-size)?

We will use the following variables, namely:
a = the smallest count (or occurrence).
b = the count of the tag being computed.
c = the largest count.
w = the smallest font-size.
x = the font-size for the tag. It is the unknown.
y = the largest font-size.


Now let's substitute the given values to their respective variables. Assuming that we are solving for the "thanksgiving" font-size.
a = 88
b = 168
c = 211
w = 12
x = ?
y = 50

And here's the formula:

x = (b-a) (y-w)
----------- + w
(c-a)

Or to put it in one liner (using c-like syntax):

x = ( ((b-a) * (y-w)) / (c-a) ) + w

And that's it. That's the formula. You might be wondering where I get that formula. Well, it's hard to explain here in words but let me still try. Using the "ratio and proportion" in Mathematics, the ratio of the distance between a and b and the distance between a and c is equated with the distance between w and x and the distance between w and y.

Or to make it simple,

b-a x-w
----- = -----
c-a y-w

Let's now continue computing the font-size for the thanksgiving. By substituting the values to the equation above, we will have...

x = ( ((168-88) * (50-12)) / (211-88) ) + 12
x = 36.715446
x = 37

The thanksgiving tag should have 37px font-size in the tag cloud. Try computing for the rest of the tag. You will get:
birthday = 29px
christmas = 18px
valentines = 50px
thanksgiving = 37px
liberation = 12px
halo ween = 20px
new year = 28px

--End

tip: When using Java, operate on float data type, not integer.

23 November, 2007

OpenSolaris Not Yet Ready For My Laptop

Recently, I blogged about the disaster that I experienced regarding the ext3 file system and concluded that I would want to try the Zettabyte File System on Solaris.

I have an HP Pavilion dv5600 laptop.

I downloaded and installed the Solaris Express Developer Edition on my laptop. This Solaris version is based on breakthrough development of the OpenSolaris operating system. But after trying it on my laptop, I learned that it doesn't support my HP laptop hardware yet. The Indiana OpenSolaris Developer Preview did not give me a good luck either.

Another operating systems that make use of the ZFS are the FreeBSD Alpha and the Mac OS X Leopard. I don't want to use FreeBSD which has the ZFS because it isn't stable yet. Mac OS X Leopard is too expensive for me; I can't afford it.

So, what now? I'm back to Linux. Linux is the best open source OS as far as the firmware/drivers is concerned. My Ubuntu Linux runs almost perfect on my HP Pavilion dv6500 laptop. But as soon as OpenSolaris is able to run on my laptop, I will always be willing to switch to it, only because of ZFS that I've been wanting to experience.

20 November, 2007

The hunger for a bite of Zettabyte

I've been using ext3 as my filesystem of choice since I touched Linux sometime in 2004. And I've never been very disappointed until now. There was a time that I thought that reiserfs is better than ext3 so I switched, but this reiserfs betrayed me when it's journal got corrupted and I could no longer recover my data. Badly irritated, I switch back to ext3 and I'd been so happy using this ext3 until yesterday.

What happened yesterday? I booted Linux, entered gnome, opened nautilus, then my computer hanged. I couldn't use the keyboard and mouse so I forced a shutdown to my computer by pressing my computer's power off button. And surprised! My computer don't boot anymore. GRUB was giving error 17.

What did I do? I downloaded Ubuntu Live-cd and patiently waited for 8 hours for the download to finish. Then I borrowed my friend's laptop to burn the ubuntu iso image. I burned it and booted it and run it. And from the running live-cd, I googled for troubleshooting the GRUB error 17 and found out that the ext3 partition of my hard disk no longer contains a readable data. I googled again and again but found no appropriate solution. When I realized that I've already spent half of a day googling and reading solutions but finding no real answer, my last resort was to follow someone's suggestion and that is to use the fsck utility.

I run the dangerous fsck....

done...

I rebooted but now GRUB gives error 15. I booted from the live-cd again to see what's inside the "rescued ext3 partition". To my surprised, the ext3 was then able to be mounted. I thought my problem was already fixed. But when I looked inside the partition. I was shocked. It only contains a single "Lost+Found" folder that contain full of directories and files with names I can not understand. Where's my data!!!! I shouted.

After all the hassles and boring journey of trying to rescue the corrupted
ext3 filesystem, my reward is a "lost of data" award. I lost the files that I've been keeping since 2001. I lost the project files that I was currently working on. Can you imagine how miserable life ext3 gave me?

Dismayed, breathing deeply and with crossed eye brows, my only solution was to reinstall a new OS. The data that I lost from that corrupted ext3 is a big pain in the a$$.

"If I'm going to install a new OS, what would it be?" I asked myself. I remembered there is a filesystem that they call as the Zettabyte File System or ZFS. I think it is the best file system available today. It is the default filesystem of any opensolaris distribution. It's maybe the right time to try opensolaris and take advantage of its ZFS goodness. What I want about ZFS is its ability to self heal and its defensive and careful approach to managing data. I hope, with ZFS, I will never experience again the same trouble that ext3 gave me.