Create algorithm to condense information

Our client has a feature which scrapes external url text/imagedata. However on many cases there is alot of headers and footers being extracted and in many cases there is paragraphs of full text being extracted. We need a smart application that recognizes as such and can highlight paragraphs and reduce text scraped without losing the main principal. For example the application scrapes a wikipedia page including header footer of wikipedia and thousands of words within that article. So we need an application that can simplify text and cut down on articles using a smart algorithm. So condense information on a page without losing touch of what the information is about. Please only bid if you can deliver. If you have created something similar in the past please send a PMB. I look forward to your bids. Deliver within 2 days. Thanks.