Expansion of Project Tiger to GLOW (Growing Local Language Content on Wikipedia)Edit
What is it?Edit
In 2017, the Wikimedia Foundation and Google launched Project Tiger, an initiative to support editors in expanding and improving articles in underrepresented languages on Wikipedia. Working in close collaboration with the Centre for Internet and Society (CIS), Wikimedia India chapter (WMIN), and local volunteers, we hosted a pilot program to increase locally relevant content in 12 Indic languages.
Google provided Chromebooks and internet access to support volunteer editors with content creation as well as insights into popular search topics on Google for which no or limited local language content exists on Wikipedia. Through a three month writing competition, volunteers created nearly 4,500 new Wikipedia articles across 12 languages, nearly double the initial benchmarks for the project. After positive feedback and discussion with our volunteer communities, including at the annual Wikimania conference in Cape Town last year, we will be expanding this pilot into something we’re calling the “GLOW” program—Growing Local Language Content on Wikipedia. We will be working with Wikimedia affiliates and volunteers in India, Indonesia, Mexico, the Middle East and North Africa, and Nigeria to include 10 more languages.
You can also find more information about this in the Wikimedia Foundation blog announcement about the expansion of the program.
How did you work with Wikimedia communities on Project Tiger?Edit
Project Tiger was the initial pilot project; the expanded program is referred to as GLOW (Growing Local Language Content on Wikipedia). We launched Project Tiger after Indic communities expressed interest in wanting support to create more content and bring in new editors to smaller language Wikipedias. During the project development phase, the community clarified that lack of resources was one factor contributing to a slower pace of article creation on their Wikis; a community needs assessment conducted by CIS as part of Project Tiger confirmed this need.
Project Tiger was designed to support editors in expanding and improving articles in underrepresented languages on Wikipedia. Google supported the project by providing Chromebooks and internet access support to editors, as well as insights into popular search topics on Google for which no or limited local language content exists on Wikipedia. Through a three month writing competition, volunteers created nearly 4,500 new Wikipedia articles across 12 languages, nearly double the initial benchmarks for the project. The communities that won the contest (Punjab and Tamil) were awarded a four-day skills development training after the contest concluded.
How did you select the countries that will be involved in the GLOW program?Edit
Taking into account the priority locales between Google’s Next Billion Users markets and Wikimedia’s anticipated growth regions, we chose to focus on India, Mexico, Nigeria, Indonesia and the Middle East and North Africa region based on the need for content in the primary language of these countries/region, and the existence of active Wikimedia communities. Though we have proposed "countries", the key issue here is the language wiki - the countries are the organizing spaces where program activities will take place.
What are the 10 languages that will be included as part of the Project Tiger expansion?Edit
The 10 languages that will be supported include Bengali, Hindi, Punjabi, Tamil, Telugu, Malayalam, Spanish, Bahasa Indonesia, Arabic, and English. Our decision to focus on these languages is based on factors such as:
- whether the language's Wikipedia is lacking in content,
- the number of people using that language,
- the growth potential for the language's Wikipedia, and
- interest and capacity of each of those Wikipedia communities to participate in the program.
Did you consult the Wikipedia community before launching this initiative?Edit
During the pilot and after we shared the initial results of the pilot at Wikimania 2018, we heard other communities were interested in doing a similar program with their editors. As we scoped out the approach and anticipated outcomes of the program, we engaged communities to get their input and feedback on approach and program benchmarks, determined their interests and capacities, and developed tentative outlines. These consultations are continuing now as we prepare to launch in each community/country. Each community is empowered to design and implement a program approach in a way that works for their context.
How would you describe the relationship between Google and Wikimedia?Edit
Google and Wikimedia share a commitment to making information more accessible to people around the world and ensuring that the information on the web reflects the diversity of its users.
More information about our most recent work together on the GLOW program and other ways we’re collaborating can be found in this announcement on the Wikimedia Foundation website.
Google partnership announcementEdit
What does the partnership announcement cover?Edit
The announcement on January 22, 2019, covered the following partnership initiatives with Google:
- Project GLOW and its pilot in India
- The use of the Google Translation engine in the Content translation tool
- Access to Google Cloud Custom Search API and Google Cloud Vision API at no cost
Does Google receive data or other personal information through sharing the Custom Search API and Cloud Vision API?Edit
The tools that use these two APIs (Earwig’s copyvio detection and Wikisource OCR) have been using them since mid-2016. With these new partnerships, Google will now be providing access to those same APIs at no cost to the Wikimedia Foundation. No changes to the tools were made, and no user or other personal data will be shared with Google to continue using these APIs.
How do the tools that use Google APIs (Content Translation, Earwig's copyvio detection, and Wikisource OCR) protect the privacy of editors who use these tools?Edit
To protect the privacy of the user, each of these tools uses the Wikimedia servers as an intermediary between the user's browser and access to Google. This design ensures that the only connection information that the Google servers can detect are the requests coming from the Wikimedia servers, and not the individual editor's device or their IP address.
- Content translation/Machine Translation/Google Translate - Machine translation support for Google Translate
- Supporting Indian Language Wikipedias Program - Wikimedia Foundation and Google partnership working with the Centre for Internet and Society (CIS), Wikimedia India chapter (WMIN) and user groups in a pilot a program encouraging Wikipedia communities to create locally relevant and high-quality content in Indian languages. AKA Project Tiger
- Frequently asked questions about the Content Translation tools
- Wikisource#Tool to use Google OCRs in Indic language Wikisource - 2015 Community Wishlist Survey proposal for using Google OCR in Indic language Wikisource
- m:Community_Tech/Google_OCR_for_Indic_language_Wikisources/notes - Notes on Community Wishlist Survey item: "Tool to use Google OCRs in Indic language Wikisource"
- Related Phabricator task
- Tracking Phabricator task for adding more machine translation services