3 ms·
That one surprised me the most. Message at http://code.google.com/apis/language/translate/overview.html http://code.google.com/apis/language/translate/overview.
by jot 15y ago
That one surprised me the most. Message at http://code.google.com/apis/language/translate/overview.html http://code.google.com/apis/language/translate/overview.html:
"The Google Translate API has been officially deprecated as of May 26, 2011. Due to the substantial economic burden caused by extensive abuse, the number of requests you may make per day will be limited and the API will be shut off completely on December 1, 2011. For website translations, we encourage you to use the Google Translate Element."
Surely "abuse" means they built something people want. Sad that they don't seem to think it would be worth their while to charge for it.
- wslh 15y agoThe "abuse" is easily organized charging for the service and including a TOS. I hope they realize that there is a business adding APIs and offering a self service console like adwords. I call it: "Google NoAPIs". And I even wrote an article about it at: http://blog.databigbang.com/google-search-no-api/ http://blog.databigbang.com/google-search-no-api/
- saurik 15y agoOr at /least/ allowing people to cache the resulting data... it was downright stupid that the Translate API's terms of service demanded that, even if a hundred thousand people all wanted a single sentence translated, you could not save the result of the single call and reuse it: you had to have all hundred thousand of those people make their own request.
- skymt 15y agoPerhaps "abuse" means toys like this: http://www.translationparty.com/ http://www.translationparty.com/
- andymurd 15y agoI think the "abuse" they are referring to is the common practise among blackhats of using Google Translate to spin stolen content to make it unique. Translating English -> $LANGUAGE -> English makes for a cheap, low grade article rewriter.
- chalst 15y agoNow Google Translate could pass information to their search quality team so that they can recognise this stolen content. Might be computationally/resource expensive, though.