Hey, here's an idea for an extension to your really cool idea if you want to make this a kick-ass commercial service:
1. Break licenses down into titles and paragraph sections and run an MD5 or SHA hash on each section to get a "fingerprint" of that section of the TOS.
2. Allow users to check off or redline specific sections of licenses they come across. If a license is "all green" it's approved for use by you.
3. Allow organizations and groups of individuals to share these green and redlining sections of licenses.
4. If a new license is encountered, you can then show "similar licenses you have accepted or rejected" — especially if a section is word-for-word the same.
5. If you really get into ML training you can do this not just for identical but similar sections of license acceptance/rejection.
I do love seeing the exact diffs. It's a cool tool for legal and IT teams trying to get their hands on all the clickthrough licensing they face. Let me know if you like the above ideas and, if you use them, all in return I'd ask is just credit me by name, perpetually free and royalty free, somewhere in the code for the suggestions.
1. Break licenses down into titles and paragraph sections and run an MD5 or SHA hash on each section to get a "fingerprint" of that section of the TOS.
2. Allow users to check off or redline specific sections of licenses they come across. If a license is "all green" it's approved for use by you.
3. Allow organizations and groups of individuals to share these green and redlining sections of licenses.
4. If a new license is encountered, you can then show "similar licenses you have accepted or rejected" — especially if a section is word-for-word the same.
5. If you really get into ML training you can do this not just for identical but similar sections of license acceptance/rejection.
I do love seeing the exact diffs. It's a cool tool for legal and IT teams trying to get their hands on all the clickthrough licensing they face. Let me know if you like the above ideas and, if you use them, all in return I'd ask is just credit me by name, perpetually free and royalty free, somewhere in the code for the suggestions.