I was wondering how to create a tag cloud based on post contents

This site is best viewed in a modern browser with JavaScript enabled.

I was wondering how to create a tag cloud based on post contents

MyDarkPassenger

I'm trying to think of how to create a tag cloud based on the content of threads/posts. I haven't created one yet but I believe the best way to create one would be to take each thread and split the text based on space. I would then cycle through the words and remove any word under say 3 characters. I would then iterate over the remaining words removing noise words (which I would put in noise array) and any words that weren't in that array I'd add to a 2 dimension array called tags (with a count). If the word exists in the array I'd increment one to the value in the array, if not I'd add the result to tags. Afterwards I'd have the ability to sort by name or count. However, I don't know if this is the most efficient way to do this. Does anyone know of a better method?

root

This is more of a server-side issue like PHP + SQL

MyDarkPassenger

\\.\;1314011 wrote:
This is more of a server-side issue like PHP + SQL

Yes and no. I tend to like to put as much as I can on the client side for overall performance. Spread the workload. SQL would definitely make it easier, however I can't see many reasons not to make some use of client side.

root

tag clouds are server side and not client side and once created add no additional work load to a server, only thing the server has to do is make a simple and quick SQL query.

You do not keep large numbers of tags, your tag cloud is meant to be what is currently trending.

MyDarkPassenger

\\.\;1314061 wrote:
tag clouds are server side and not client side and once created add no additional work load to a server, only thing the server has to do is make a simple and quick SQL query.

You do not keep large numbers of tags, your tag cloud is meant to be what is currently trending.

Tag cloud was probably the wrong word. Here's what I'm working on. I'm trying to make an RSS reader, like Google Reader. What I plan to have is the ability to use regular SQL or full text searches (obviously) but also have filtering capabilities based off the returned results. Here's the difference. When you do a search for foobar you do a SQL search against the site. The results are returned. You can then start typing tag1 or tag2 and it will only show the results from the search that contain the tag. Clearing them out just shows the full results. There's no server involvement. I'd also like a side tag cloud that shows the tags used on the page and you can click on them to filter the results. I've seen some clouds on other sites like this. This can give the user very fast results while also putting less stress on the server.

MyDarkPassenger

I think I just noticed some faulty logic on my end. The server results can easily do the tag cloud. The cloud can just then filter the results.