In our series of portraits and perspectives on Localization Business Innovation, Francis Tsang, director of globalization at Adobe Systems Inc., explains the opportunities opened by using 'crowdsourcing' as a localization solution, and the language data sharing initiative as a key force for supporting the next level of globalization
Put simply, the economics of crowdsourcing involves leveraging a user community's skills and interest to deliver a specific service to the community and a (content) producer. The pay-off is a win-win: the crowd gains access to knowledge, while the producer expands its markets and builds greater product equity. Language data sharing will trigger a genuine ecosystem of services, lower the bariers of entry for smaller players.
How can the 'crowd' as a resource be best harnessed for the translation and localization industry?
Francis Tsang: "My top level globalization agenda is to reduce the incremental workflow and management complexity to support additional languages for our products. There is far too much content that needs translating, so we have to decide which is high priority. I believe that if we allow a community of users to start making those choices, then we gain a powerful insight into which parts of our products are most important."
"Our ultimate goal is to discover what is the optimal user experience for our products and then project this at a level of service. By first crowdsourcing our translation, we as product makers can achieve this kind of insight. From a go-to-market perspective, the virtue of crowdsourcing is that it allows us to build products and markets at the same time. We can rapidly build an ecosystem around a set of products, offering more content, training materials, books on APIs and so on in more languages with the support of these early adopter users in new locales. We would never be able to generate this kind of momentum by using the traditional localization model."
Isn't there a risk of using the community as a cheap resource?
Francis Tsang: "If you simply try and get the crowd to help your company, you will end up with a severe management problem and lose out. But if you use crowdsourcing to grow communities around products, they will take control and your job is then to give them the right kind of sustained support. For me, crowdsourcing is based on the concept of user-driven services. Companies like Adobe and Sun Microsystems use naturally-emerging crowds with specialist knowledge about products who want more product content in their own languages. This way we can localize product-related content for languages which may not be high priority, but which nevertheless help grow our markets."
How does this affect issues such as quality and service levels?
Francis Tsang: "You can not apply explicit standards or service level agreements, since they cannot be enforced. Quality will emerge from the commitment of the crowd to the value of their work. Our job is to provide the kind of terminology, style guides and file support (but probably not translation memories) that enhances the potential for high quality translation. In our experience, the only real problem is maintaining consistency from one release to the next. You can get great people for one version but they are not around for the next. So the key to good quality is therefore to have a very large community of the product or technology in the marketplace. And the ideal crowdsourcing task is the kind of product content found in a version 1 or 1.5, not content for a mature software product."
Who can use crowdsourcing most effectively?
Francis Tsang: "Obviously it is not a one-size-fits-all solution. It requires an open-source mindset, and depends on whether the target is rapid international growth or growing a broad community of faithful users. If the latter, then the producer must provide the tools to support the community and motivate them. Actors should not think of crowdsourcing as just an alternative model: for them it is the only show in town. And you should explain why. To provide that support you need to restructure your whole workflow to eliminate the risk of errors and their accompanying overheads, and organize everything with an external supplier in mind. The aim is to allow anyone in the crowd to take on any work at any time in a continuous flow model, regardless of whether they are inside or outside that product cycle. This means providing the right management support. Naturally it depends on the type of community involved (some are self-driven, while others are more top-down) and in some cases a vendor may be the best agent to manage it for you."
How about the role of automation in a community perspective?
Francis Tsang: "If our aim is to work with communities to help them grow and to expand our markets, then we need to expand the role of content self-service to find out what our customers want. The issue is not just volume but having some way to rank the content users need in a relevant way. This could even influence authoring practice, for example. So I am in favor of any kind of automation that facilitates faster translation turn-around so that user can access what they want. MT for gisting, for example, can help in determining what is important for a user. But the focus must always be on delivering a new user experience, not just on solving a one-off resource problem. This is the essence of building a product and a market simultaneously."
Will the emergence of very large language data resources impact this model?
Francis Tsang: "The industry is now reaching a new plateau. We need appropriate business models to handle new requirements. I see three critical benefits from the TAUS initaitive to open up language data sharing facilities. One, it will accelerate time-to-market to a matter of minutes. In future, anyone should be able to draw on data to speed up translation, not just five or six large vendors. One obvious application here will be to provide better training for MT systems, for example. Two, more data sharing will help build industry-wide consistency in terms and style. This is not just QA for QA's sake. It will ultimately help end users master information more effectively, and provide a common basis for understanding content. Three, a data sharing facility will trigger a genuine ecosystem of services. It will lower the barriers of entry for smaller players and perhaps cause a structural change in attitudes that will help us address the future model more realistically. Currently vendors are understandably preoccupied with making money. But this is not part of the long-term solution to the translation issue. We need to foster new skills, work through communities, experiment with business models. Data sharing will be one of the key forces for supporting the next level of globalization."
上一篇：
发表、查看更多关于该信息的评论
将本信息发给好友
打印本页