I used to work at a publishing services company editing the text of books. If I was looking at a page with an editor's mark, and I wanted to find that place in the file, I'd look for a nearby unusual word or phrase and search for that in the file.
So my suggestion is to focus on the parts of each text that make them most unique. Take a word histogram of the original page text and look for the least common words in the quote. From there, you can look nearby in the original page and see if other low frequency words are nearby.
I can't tell from your problem description whether you're more concerned with false positives or false negatives. I think the algorithm I'm describing could be fairly tricky, and probably hard to tune to a particular false detection rate. Still, depending on the nature of the texts you have, it might pick up things other matching methods miss.
In reply to Re: Verifying a quote matches (closely enough) a source URI
by kyle
in thread Verifying a quote matches (closely enough) a source URI
by Your Mother
| For: | Use: | ||
| & | & | ||
| < | < | ||
| > | > | ||
| [ | [ | ||
| ] | ] |