Keyword Extraction Tool

[Robber]

I am trying to develop some programming skills by building some tools to help with website building. I don't normally pay too much attention to the meta keywords tag but I feel I should make the effort, so I thought Id write a programme to build the keywords tag for me. The tool Ive built basically takes all the text in the body of a webpage and gets rid of all the duplicate words and words like "a, and, or if" etc etc, you can add your own words to the list too, it then presents you with a meta keywords tag. The basic idea is that it should just present you with important words that are contained in your page.

<snip>

I know of a quite a few things that need sorting but Im sure you guys can find plenty more!!

Cheers

[edited by: Xoc at 4:47 pm (utc) on June 23, 2002]

[Marcia]

>takes all the text in the body of a webpage and gets rid of all the duplicate words and words like "a, and, or if" etc etc, you can add your own words to the list too, it then presents you with a meta keywords tag. The basic idea is that it should just present you with important words that are contained in your page.

Robber, programming is a challenge and a lot of fun, and good tools are very useful if they're geared to the final result.

A lot of people put all the words they can think of relating to the page into the meta keywords tag, but the general consensus is that a page should ideally target one main phrase. So having a lot of words in that tag can dilute the value of the phrase being targeted, and could possible emphasize phrases other than what the webmaster intended, depending on the density.

Take a look at Brett's Keyword Density Analyzer [searchengineworld.com]. I generally run all pages through that, as well as sometimes opening them in a piece of software, stripping the html and doing a density check.

On one site recently, I found that the main page was targeted to a phrase that gets a couple of hundred searches a month per Overture and generates no revenue instead of one that has many thousands of searches and represents the product being sold. I pulled the meta tags and replaced with them with new ones including only the few words being targeted and also pulled about half the text from the page.

Probably a density analyzer that would take an additional step and generate a meta keywords tag based on a few selected words would be a tremendous time-saver for people. That could be multi-step, but would be very handy.

Any other types of tools you're thinking of generating in addition?

[Robber]

Hi all,

thanks for the feedback.

Marcia, your thoughts are spot on there, I actually pay very little attention to the keywords tag, I don't really think the time spent on crafting the keywords tag provides a good return on time spent (generally speaking!).

It was this feeling, along with the desire to learn programming that motivated me to try and write this little application. Its fundamental premise is that a keywords tag should only include words that are actually in the body of the page - therefore taking the body text and stripping the rest out seemed like a fair starting point.

<snip>

Cheers all.

[edited by: Xoc at 4:49 pm (utc) on June 23, 2002]