Universal article parser. What is the best way to implement the algorithm?

Question

PHP

There are no problems with php itself, as well as with regular expressions.
The idea is this:
We take an average article site (namely, 5 links to articles).
Next, you need to compare these 5 pages with each other, presumably the html code before and after the article is static, otherwise we will consider what changes as the content we need. (the block "similar articles" can get in, but for now it doesn't matter)
Help me figure out how to compare several html pages in order to understand which part of them remains unchanged.