Yeeyan

Yeeyan (http://www.yeeyan.com) is the largest open translation community in China. To date (June 30, 2009), it has more than 90,000 registered users, with about 5,000 community translators who have published nearly 30,000 translations on the site.

Recently The Guardian (http://www.guardian.co.uk), collaborating with Yeeyan, launched their ground-breaking project - The Guardian's Chinese UGC (User Generated Content) site (http://guardian.yeeyan.com)  on May 18, 2009.

Discover, Translate and Read the Internet Beyond Your Language

"Discover, translate and read the Internet beyond your language" is Yeeyan's slogan. We are doing it by three core processes: recommending, translating, and publishing. Each process serves different needs.

The recommending process is to allow users to recommend interesting articles to the community. It is also an effective means for the site to organize the community to produce desired content. "Crowd recommending", which allows users to vote for recommendations, was one of the earliest features but removed later due to inactivity. Now Yeeyan is going to output filtered recommendations to mini-blogs, such as twitter. Together with the mini-blog API, we are expecting "crowd recommending" to become an often-used feature and help the community to identify demands.

In the translating process, the tools for translation work are actually a small part: the key task is to provide a reliable on-line rich text editor to users so that they can type their translation in the editor or simply copy/paste translations from elsewhere. The real heavy duty work of the translating process is "quality control". We have developed a tool to allow users to indicate translation errors by "in-line comments". Error rate is calculated based on the number and type of valid errors, which gives the quantitative metric of a translation. The formula is learned from the traditional translation industry. Other quality control tools include version control and translator evaluation. Quality control not only guarantees service quality delivered to clients, but also provides translators with helpful and meaningful rewards for enhancing their comprehension skills.

The publishing process is responsible for organizing and displaying translations, i.e., for reading purposes. A hybrid mechanism that combines a Digg-like computer algorithm and community editors' choices recommends "top" translations to readers. Selected translations are also output to mini-blogs. The publishing process will also provide service interfaces, which may include customized subscription, and most importantly, a licensing/purchasing service.

Community-Based Enterprise

No matter how technology can help, translation itself is a labor intensive work, or in another words, humans are always at the center of a translation. This is particularly true for "open translation" when a community (or crowd) is involved.

Yeeyan is step by step building a hierarchy in the community. We have editors and proofreaders from the community to do daily recommendations and proofreads; we even have a column editor from the community to take a major responsibility for The Guardian's Chinese UGC site. Currently, above 50% of Yeeyan's gross income is rewarded to the community, even though the company is still fighting for its survival.

The current structure is still centralized, which means Yeeyan has a direct control on all commercial projects. For non-commercial projects, Yeeyan provides a wiki platform (http://pro.yeeyan.com)  to allow users to play on their own. A very successful project organized by users is the translation of materials from the Leukemia and Lymphoma Society (http://pro.yeeyan.com/wiki/%E8%AF%91%E7%88%B1). The project's name means "Love by Translation". It has finally gained sponsorship from Google's charity fund.

The centralized structure is just a middle stage. By transplanting project mechanisms from inside the company into the community through product development and training the community, we are aiming at building a decentralized, self-organized community, which allows users to run commercial projects on their own.

Yeeyan has been complained for ineffective communication with the community for quite a long time. Active users submitted suggestions, opinions and bug reports to Yeeyan's BBS (http://bbs.yeeyan.com) but rarely received prompt feedbacks. Yeeyan's blog (http://blog.yeeyan.com)  was not updated regurlarly. Even product releases were not announced and explained to the community in a timely manner. To improve this, Yeeyan has emphasized within team that product-related posts such as bug reports on BBS must be replied as soon as possible. Yeeyan is also considering to start biweekly addresses on its blog. Future steps may include opening a bug-tracking system to the community.

By doing these, Yeeyan is making its way toward a community-based enterprise.

Copyright-Related Issues

Copyright is an inevitable issue for open translation. It has been discussed and argued within the team since the first day of Yeeyan. By attending OTT09 (Open Translation Tools 2009) in Amsterdam, Yeeyan found its long-sought community and is willing to cooperate with Creative Commons to resolve copyright related issues and construct licensing platform on top of Yeeyan's UGC engine. This is the next big thing for Yeeyan.