In a recent mail to the google group regarding Google's Indic Wikipedia translation project, Dimple Batra from Google reported:
"We are in the process of closing down the wikipedia indic language translation project.
We had initiated this project to bootstrap the creation of indic content and encourage consumption in indic languages. Having accomplished the same over the past 2 years we would now like the community to continue contributing.
Please feel free to reach out to us if you have other proposals on how we can create more local content in India"
***
Please let us know how the different Indic Wikipedia communities involved in this project are going to handle this.
As far as Tamil Wikipedia is concerned:
1000+ articles created using this project went through few rounds of corrections by both Wikipedians and the paid translators. About 50% of them were found to be of acceptable quality. We selected few translators who were doing a better job and a test case of 25 articles were assigned based on which further articles should be written. But, the collaboration between Tamil Wikipedia and Google lost its momentum after this stage.
Frankly, Tamil Wikipedians were exhausted in this year long project with more than 20 Wikipedians contributing for this. The good will between Google - Translators - Wikipedians did not improve as the translators didn't show interest in improving their previous articles any further until new articles are assigned to them. (Which means continuous employment for them and profit for the companies involved). Later, we came to know that most of the translators involved in the project resigned from their respective companies and expected a stand still in this project.
In this situation, the above reply from Google came for a query posted by Arjuna Rao Chavala regarding the project's status in Telugu Wikipedia.
We haven't decided yet on the next course of action on the Google created articles. My expectation is that they will be treated as regular articles and the community will try to work on them. In case of very badly written articles, part of the article may be moved to talk page and moved back again after they are corrected. However, this will really be a time consuming exercise for sure.
My personal opinion is that I am very much disappointed. Google's intention was good but execution left very much to desire. It didn't show interest to talk to the community before starting and before ending the project abruptly on its own.
Regards,
Ravi
Ravi, What Google did can pretty much be summarized as a 'Half hearted attempt'. However, I had posted last week, a link to Microsoft's API and transliteration tools. Personally, I find it better than Google. The only catch is that Microsoft requires you to link back to them. -- Regards, Srikanth
On 5 June 2011 21:44, Ravishankar ravidreams@gmail.com wrote:
In a recent mail to the google group regarding Google's Indic Wikipedia translation project, Dimple Batra from Google reported:
"We are in the process of closing down the wikipedia indic language translation project.
We had initiated this project to bootstrap the creation of indic content and encourage consumption in indic languages. Having accomplished the same over the past 2 years we would now like the community to continue contributing.
Please feel free to reach out to us if you have other proposals on how we can create more local content in India"
Please let us know how the different Indic Wikipedia communities involved in this project are going to handle this.
As far as Tamil Wikipedia is concerned:
1000+ articles created using this project went through few rounds of corrections by both Wikipedians and the paid translators. About 50% of them were found to be of acceptable quality. We selected few translators who were doing a better job and a test case of 25 articles were assigned based on which further articles should be written. But, the collaboration between Tamil Wikipedia and Google lost its momentum after this stage.
Frankly, Tamil Wikipedians were exhausted in this year long project with more than 20 Wikipedians contributing for this. The good will between Google
- Translators - Wikipedians did not improve as the translators didn't show
interest in improving their previous articles any further until new articles are assigned to them. (Which means continuous employment for them and profit for the companies involved). Later, we came to know that most of the translators involved in the project resigned from their respective companies and expected a stand still in this project.
In this situation, the above reply from Google came for a query posted by Arjuna Rao Chavala regarding the project's status in Telugu Wikipedia.
We haven't decided yet on the next course of action on the Google created articles. My expectation is that they will be treated as regular articles and the community will try to work on them. In case of very badly written articles, part of the article may be moved to talk page and moved back again after they are corrected. However, this will really be a time consuming exercise for sure.
My personal opinion is that I am very much disappointed. Google's intention was good but execution left very much to desire. It didn't show interest to talk to the community before starting and before ending the project abruptly on its own.
Regards,
Ravi
Wikimediaindia-l mailing list Wikimediaindia-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikimediaindia-l
Hi Srikanth,
I have found the tools by both Google and Microsoft to be just OK. But, I haven't seen one regular Wikipedian using them for contributing. I have found that directly working in Wikipedia is much user friendly than using their tools.
But the problem in Google project's case is not about the tool. Google funded lots of paid translators to manually translate articles using their tools. This created lot of operational problems in the respective Wikipedia communities.
More details can be found at this presentation at http://bit.ly/adRtak
Whereas, Microsoft just stopped with the creation of tools and did not seed any article directly in mass scale.
Ravi
On Sun, Jun 5, 2011 at 10:14 PM, Srikanth Ramakrishnan < rsrikanth05@gmail.com> wrote:
Ravi, What Google did can pretty much be summarized as a 'Half hearted attempt'. However, I had posted last week, a link to Microsoft's API and transliteration tools. Personally, I find it better than Google. The only catch is that Microsoft requires you to link back to them. -- Regards, Srikanth
Thanks for sending this, Ravi. I am wondering how much this has to do with them also shutting down their translation and transliteration APIs as recently announced and how much this has to do with them having achieved what they set out to do - not as far as Wikipedia is concerned but their internal metrics for the project.
A larger question I am trying to parse - what have we learned from it in terms of partnering with organizations that also have commercial goals?
Thank you.
Best,
Gautam ________ http://social.prathambooks.org/
On 5 June 2011 21:44, Ravishankar ravidreams@gmail.com wrote:
In a recent mail to the google group regarding Google's Indic Wikipedia translation project, Dimple Batra from Google reported:
"We are in the process of closing down the wikipedia indic language translation project. We had initiated this project to bootstrap the creation of indic content and encourage consumption in indic languages. Having accomplished the same over the past 2 years we would now like the community to continue contributing. Please feel free to reach out to us if you have other proposals on how we can create more local content in India"
Please let us know how the different Indic Wikipedia communities involved in this project are going to handle this.
As far as Tamil Wikipedia is concerned:
1000+ articles created using this project went through few rounds of corrections by both Wikipedians and the paid translators. About 50% of them were found to be of acceptable quality. We selected few translators who were doing a better job and a test case of 25 articles were assigned based on which further articles should be written. But, the collaboration between Tamil Wikipedia and Google lost its momentum after this stage.
Frankly, Tamil Wikipedians were exhausted in this year long project with more than 20 Wikipedians contributing for this. The good will between Google
- Translators - Wikipedians did not improve as the translators didn't show
interest in improving their previous articles any further until new articles are assigned to them. (Which means continuous employment for them and profit for the companies involved). Later, we came to know that most of the translators involved in the project resigned from their respective companies and expected a stand still in this project.
In this situation, the above reply from Google came for a query posted by Arjuna Rao Chavala regarding the project's status in Telugu Wikipedia.
We haven't decided yet on the next course of action on the Google created articles. My expectation is that they will be treated as regular articles and the community will try to work on them. In case of very badly written articles, part of the article may be moved to talk page and moved back again after they are corrected. However, this will really be a time consuming exercise for sure.
My personal opinion is that I am very much disappointed. Google's intention was good but execution left very much to desire. It didn't show interest to talk to the community before starting and before ending the project abruptly on its own.
Regards,
Ravi
Wikimediaindia-l mailing list Wikimediaindia-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikimediaindia-l
Hi Gautam,
I am not sure if this is related with them closing down other Indic related projects.
Two guesses:
1. They might have started these projects with a goal of collecting some data and they might have met it.
2. They might have found the Indic market less profitable at the moment and might have chosen to shut down some resources.
Google's project can hardly be called a partnership. It started the program on its own and the communities had only two choices: Block them or guide them how to do it better.
Tamil Wikipedia chose to work with them in the hope of improving this project and tried to provide a pilot model for other Wikipedias.
From my experience in this project, my personal opinion is that:
Partnerships should be mutual, transparent, well conceived and always in the best interest of Wikipedia.
*Any commercial interest in the content created will only lead to compromise of some sort. *
For example, if a book publisher wants to write a series of articles in Wikipedia and then publish it, this might seem like a good will gesture of donating articles to Wikipedia. But, the articles may not adhere to Wikipedia's MOS and may be written in a style for their target audience.
Commercial entities can best support Wikimedia by "donating" resources like tools, facilities and sponsorships to conduct events.
Ravi
On Mon, Jun 6, 2011 at 12:14 AM, Gautam John gautam@prathambooks.orgwrote:
Thanks for sending this, Ravi. I am wondering how much this has to do with them also shutting down their translation and transliteration APIs as recently announced and how much this has to do with them having achieved what they set out to do - not as far as Wikipedia is concerned but their internal metrics for the project.
A larger question I am trying to parse - what have we learned from it in terms of partnering with organizations that also have commercial goals?
Thank you.
Best,
Gautam
Thank you, Ravi. The mutual benefit is something, as you have pointed out, is really important.
Best,
Gautam ________ http://social.prathambooks.org/
On 6 June 2011 12:53, Ravishankar ravidreams@gmail.com wrote:
Hi Gautam,
I am not sure if this is related with them closing down other Indic related projects.
Two guesses:
They might have started these projects with a goal of collecting some data and they might have met it.
They might have found the Indic market less profitable at the moment and might have chosen to shut down some resources.
Google's project can hardly be called a partnership. It started the program on its own and the communities had only two choices: Block them or guide them how to do it better.
Tamil Wikipedia chose to work with them in the hope of improving this project and tried to provide a pilot model for other Wikipedias.
From my experience in this project, my personal opinion is that:
Partnerships should be mutual, transparent, well conceived and always in the best interest of Wikipedia.
Any commercial interest in the content created will only lead to compromise of some sort.
For example, if a book publisher wants to write a series of articles in Wikipedia and then publish it, this might seem like a good will gesture of donating articles to Wikipedia. But, the articles may not adhere to Wikipedia's MOS and may be written in a style for their target audience.
Commercial entities can best support Wikimedia by "donating" resources like tools, facilities and sponsorships to conduct events.
Ravi
Hi Gautam,
Sorry if I was not clear.
Without mutual benefit no entity would enter a partnership.
If a Wikipedia partners with a local museum, then it is win-win as the Wikipedia will get more content and the museum's collection is digitised.
But I couldn't imagine lots of commercial cases like this where there will be mutual benefit.
Ravi
On Mon, Jun 6, 2011 at 12:55 PM, Gautam John gautam@prathambooks.orgwrote:
Thank you, Ravi. The mutual benefit is something, as you have pointed out, is really important.
Best,
Gautam
On 6 June 2011 13:08, Ravishankar ravidreams@gmail.com wrote:
Without mutual benefit no entity would enter a partnership. But I couldn't imagine lots of commercial cases like this where there will be mutual benefit.
Indeed. And that we, as Wikipedians, need to evaluate possible partnerships from what *we* would get from it and not only because it is something that is "free" to us?
Thank you.
Best,
Gautam ________ http://social.prathambooks.org/
Ravi, Perhaps this tool would be handy? http://en.wikipedia.org/wiki/User:Keymanweb/Keymanweb
On 6 June 2011 13:11, Gautam John gautam@prathambooks.org wrote:
On 6 June 2011 13:08, Ravishankar ravidreams@gmail.com wrote:
Without mutual benefit no entity would enter a partnership. But I couldn't imagine lots of commercial cases like this where there
will
be mutual benefit.
Indeed. And that we, as Wikipedians, need to evaluate possible partnerships from what *we* would get from it and not only because it is something that is "free" to us?
Thank you.
Best,
Gautam ________ http://social.prathambooks.org/
Wikimediaindia-l mailing list Wikimediaindia-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikimediaindia-l
On Tue, Jun 21, 2011 at 23:26, Gautam John gautam@prathambooks.org wrote:
Just saw this: http://googleblog.blogspot.com/2011/06/google-translate-welcomes-you-to-indi...
So they thought they had enough training for the machines with Wikipedia translation feeding a good chunk of data to the machine in these 5 languages. Unlike Hindi when it was launched, the quality of these(atleast I can vouch for Tamil) is *really* bad, worse than Hindi.
<Joke> Google translate knows politics too! Try these in translate.google.com முதல்வர் ஜெயலலிதா முதல்வர் கருணாநிதி </Joke>
Regards Srikanth.L
On 21 June 2011 23:46, Srikanth Lakshmanan srik.lak@gmail.com wrote:
So they thought they had enough training for the machines with Wikipedia translation feeding a good chunk of data to the machine in these 5 languages.
No wonder they say "five new experimental alpha languages"!
Thank you.
Best,
Gautam ________ http://social.prathambooks.org/
In fact it affected the community growth also. Allowing this type paid editing is one of the bad things happened in Indic wikis recently.
On Tue, Jun 21, 2011 at 11:52 PM, Gautam John gautam@prathambooks.orgwrote:
On 21 June 2011 23:46, Srikanth Lakshmanan srik.lak@gmail.com wrote:
So they thought they had enough training for the machines with Wikipedia translation feeding a good chunk of data to the machine in these 5 languages.
No wonder they say "five new experimental alpha languages"!
Thank you.
Best,
Gautam ________ http://social.prathambooks.org/
Wikimediaindia-l mailing list Wikimediaindia-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikimediaindia-l
Great, now in ta wiki we are stuck with the task of correcting a 1000 crappy large articles. No wonder they showed little interest in correcting or copy editing and always wanted to push for more new articles. They must have had an internal target of X words or so for gaining experience for the machines. Once that was done they have closed up shop.
"You can expect translations for these new alpha languages to be less fluent and include many more untranslated words"
If this is the officially admitted state of an alpha release, imagine the quality of the articles this project created using beta versions.
For Tamil wiki there was a control over the Google's project. I hope that had helped to control the damage. But think about the wikis where Google employees were having a free hand.
On Wed, Jun 22, 2011 at 9:09 AM, Bala Jeyaraman sodabottle@gmail.comwrote:
Great, now in ta wiki we are stuck with the task of correcting a 1000 crappy large articles. No wonder they showed little interest in correcting or copy editing and always wanted to push for more new articles. They must have had an internal target of X words or so for gaining experience for the machines. Once that was done they have closed up shop.
"You can expect translations for these new alpha languages to be less fluent and include many more untranslated words"
If this is the officially admitted state of an alpha release, imagine the quality of the articles this project created using beta versions.
Wikimediaindia-l mailing list Wikimediaindia-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikimediaindia-l
Hi,
I tried how translate works translating my English bloghttp://tech4society.blogspot.com/to Telugu. Given that the software is alpha, the number of untranlated words were quite few (abbreviations or words like focussing, inputting, rapidy mostly words with spelling errors). Grammatical construction of sentences is weak. For people who are not comfortable in understanding English, the service may give atleast an idea of the original text context.
While I appreciate the efforts of Google in introducing Google Translate support for english languages, considering that Indian government and organisations were not able to offer such service for Telugu and support for other languages also not available for free access (Link for English ->other languages http://tdil-dc.in/components/com_mtsystem/CommonUI/homeMT.php), I wish that they had engaged with Wikimedia community in a better way to make their contributions to Wikimedia met the expectations from the community.
Thanks Arjun
A follow up:
The alpha release has created another problem in ta wiki. People are starting to use it for creating articles and they are mostly gibberish. We have created a new speedy deletion criteria for tagging and deleting them.
On Wed, Jun 22, 2011 at 10:13 AM, Arjuna Rao Chavala < arjunaraoc@googlemail.com> wrote:
Hi,
I tried how translate works translating my English bloghttp://tech4society.blogspot.com/to Telugu. Given that the software is alpha, the number of untranlated words were quite few (abbreviations or words like focussing, inputting, rapidy mostly words with spelling errors). Grammatical construction of sentences is weak. For people who are not comfortable in understanding English, the service may give atleast an idea of the original text context.
While I appreciate the efforts of Google in introducing Google Translate support for english languages, considering that Indian government and organisations were not able to offer such service for Telugu and support for other languages also not available for free access (Link for English ->other languageshttp://tdil-dc.in/components/com_mtsystem/CommonUI/homeMT.php), I wish that they had engaged with Wikimedia community in a better way to make their contributions to Wikimedia met the expectations from the community.
Thanks Arjun
Wikimediaindia-l mailing list Wikimediaindia-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikimediaindia-l
http://blogs.wsj.com/indiarealtime/2011/06/24/how-japan-helps-google-transla...
And then they don't even thank the communities that helped. *sigh*
Thank you.
Best,
Gautam ________ http://social.prathambooks.org/
wikimediaindia-l@lists.wikimedia.org