Licensing

From My Wiki

Specific Questions on Licensing:

Contents 1 Should a CC translation license be developed? 2 is annotation a derivative work? 3 best practices for translation and licensing 3.1 not fair use 3.2 dream tool 3.3 problematic 4 Machine Translation research 4.1 who owns the © of a derivative machine translation? 5 string translation of software 5.1 what about moral rights in string? 6 automated dictionary usage

Should a CC translation license be developed?

• License to signal :please translate this work, but don't create any other derivative works
• common practice is just asking permission from author and implying their willingness to have work translated and hosted

is annotation a derivative work?

(marking verbs and parts of speech = annotation)

• is anyone else using the annotated work? --> yes, the annotated POS files will be distributed
  • private modification is still the exclusive right of the copyright holder (in any signee of Berne Convention/TRIPS...depends on commercial significance)
  • lots of companies take “private” copies of software & modify, use in house
• considered a license when you give explicit consent and accept terms (and binding contract)

best practices for translation and licensing

• should include: how to approach someone whose content you want to use without explicit license
• if believe in open content, then approach and recommend using open licenses
  • good way of getting permission: having the most permissive license
  • bad practice: don't mix licenses (esp incompatible)
  • to be responsive and take down
  • add disclaimer explaining situation and good intentions ((c) world doesn't allow us to work otherwise)

might be problem if publishing book, etc. though

• do everything you can to make it clear that you're doing everything you can (purpose and licensing of content)

“copyright is like bathing in gravel”

not fair use

• if you translate 2 out of 4 paragraphs from the work, then it's probably not fair use
• making a collection is not a derivative work itself

dream tool

• widget that inserts icon next to linked blog post explaining the license condition

problematic

• an aggregator under CC may violate licenses of source content
• hard to get a hold of blog poster (email addresses not available). therefore difficult to clear rights

Machine Translation research

• what we'd like = explicit permission
• if came to a court case over machine translation data, then could the source content could also get sued (i.e. if MT used infringing material from Global Voices)
• because working with University, might be under some protections
• how is the information stored?
  • in text documents (copy website, remove html, and get wordfreq counts but delete source from computer
• aligned text: after trained machines, would it be possible to release the data without infinging material? no, because people would want the source data to make improvements
  • has permission to align texts and releases the derivative under CC BY-SA and GPL (for the future software packaging)
  • run computations on text source, derive numbers and stats, and release with link to text source so that people can recreate
  • use Internet Archive for sources (can keep copyrighted work bc they're an archive)

who owns the © of a derivative machine translation?

• the person who pressed the button

USA Oscar Wilde Lithography (Burrow-Giles case) – Wikipedia article

string translation of software

• there should be TM under a license that can be used
• are translated strings under the licenses of the original?
• dream: people who release their software to explicitly license their strings in PD/under a license that can be used
• by releasing your strings in the PD, then it will improve the translatability of your own software (business argument)
• should be in a guidelines – best practices for compatibility (GPL + PD for strings, etc.)

what about moral rights in string?

• and non-existence of PD in some (GPL has some solutions for)
• problematic: commercial company can use GPL license not under right conditions and say that just used PD strings (could counter this by proving the method under which they acquired code)

look & feel also under ©

automated dictionary usage

• 10,000 word list look-up = probably not under © because words themselves are not copyright
• link to US case on telephone book (Feist Publications, Inc., v. Rural Telephone Service Co.)

Lie Street (fake person and fake # to catch people with infringing copyright)