Music Search Engine Theorm

[cwarn23]

Hi, I am making a music search engine and although my first version is complete, I think it's theorm could do with a bit of improvement. So I need to ask the theory of how to make a music search engine for picking key words and the description text.

 

This is how it currently goes, a domain is entered, subpages found, for each page found a function is called to index the page. This function then finds any music files within the webpage. When it finds a link with a music extension it then selects the text from 400 characters before the link till 756 characters characters after the link. It then removes html tags from that text and at the moment it places that text into a mysql database in one column, the media link in another column and the page link in the third column.

 

I have been thinking of changing that last piece so that there are 10 extra mysql columns each with a keyword and the top 10 reaccuring words are placed into those 10 fields instead of dumping the entire block in one field. But does anybody know of a way I can get the description like on google searches (theory wise - code optional) and any suggestions on how this theorm should be changed.