PHP stored content approximate duplicate match class - repost

Looking to outsource some requirements to a highly skilled PHP programmer. Upon payment all code ownership must be passed over to me. You will need to create a class to identify duplicate content. Our code will pass content and this class must include an algorithm that can identify text areas that are 1. exact duplicates and 2. approximate duplicates. I must stress we are NOT seeking a 100% direct match. Code must be smart enough to identify partial matches and be processor efficient. We should be able to define duplicate percentage for any match. So if we set content duplicate percentage to 20% then any content duplication under 20 percent for that match should not trigger alert. Any content that comes back as matching should be tagged. So if we have 10 words out of a 15 word sentence, the 10 words that are duplicate should be tagged on return so we can label and process them. The code/algorithm needs to be clever enough to work like [login to view URL] if content is a match what percentage and highlight matching content. Matching should look for keywords and order of sentences. For the right candidate, who is able to deliver quality and efficient code that satisfies the project needs, there will be more projects to come.