"Watermarking" Text Data

There are some ways to sort of watermark text: Certain letters and numbers tend to look so much like each other that replacing them in spelling inappropriate ways could be used to mark the text without appearing to. This can be used to detect the source of the text when it has been copied and pasted and to provide evidence of plagiarism. An example probably explains it better:

Love them tigers! = Love thern tigers!

Open source = 0pen source

I love you = I 1ove you

and so on. These same sorts of replacements are commonly used by spammers to try to sneak past word fi!ters for commonly spammed items. e.g. that male erection stuff. 0ver the course of a few words, or in a headline or subject, there are sort of easy to see, but in a long bit of text, you can miss them pretty easily. (go back and look at the word "Over" in the last sentence, and the more obvious "filters" in the prior).

If the copied text has YOUR specific pattern of replacements (e.g. always replace the 3rd "O" with "0" and the 1st and 4th "l" with "1" and...) then you know it came from you. A script can easily parse the text, looking for these marks and ring a bell when it finds them.

O = 0
I = 1, !
m = rn
V = \/
" = ''

Other possibilites include:

• Double spaces (very effective in HTML as only one space will be displayed)
• Depending on the characterset there are often extended symboles, some of which are close enough to the standards. E.g. i í
• Inconsistant use of abbreviations. E.g. "I live on Happy Ave. but my friend lives on Sad Street"
• Inserting text into HTML with tags that cause it to not be displayed, but that will still allow it to be copied. E.g. You see this, <font style="display:none" color="#FFFFFF">but not this, although it is copied</font>
• Replacing certain letters or symbols with a picture of that symbol. This causes the item to display, but not be copied.

  .