In today’s fast-paced world, we strive for ways to innovate, improve, and create efficiencies. The translation industry is no different. We can all agree that machine translation has changed the way we as an industry and our customers see translation. But we are yet to see any impact from crowd sourcing technologies.
We all understand that using a crowd for translation or editing is essentially work done by the mob. Where the effort that is needed to vet and QA the crowd and move a task forward can often remove the value, time, and effort balance when compared to other approaches.
The theory of using a crowd is great; but how do you monitor quality and ensure projects are progressed effectively? What is needed is a crowd that can regulate itself.
The key to the KantanStream solution is that it has a unique combination of channels, automated QES scores, and a embedded community of invigilators.
- Defines what part of the crowd can work on the segments based on their quality rating.
- Defines what segments get placed in the crowd for editing based on the translated segments QES (Quality Estimation Score).
- Defines the time the content is available on the system for the crowd to post-edit.
- QES is an automated forecasted quality score on the MT segment.
- Segments that fall below the QES score are sent for post editing.
- Segments that meet or exceed the QES score has no human intervention.
- Is the peer-to-peer process of ranking and scoring users in a group. This tool is unique as it is completely automated.
As any localization manager will tell you, you must have confidence in your process to deliver quality. In the same way you need confidence in the crowd and how it regulates itself.
Within KantanStream we have three defined roles which are automated to achieve this
- An MT post editor.
- An MT post editor reviewer.
- An invigilator (defined or elected).
Combined with an automated AI ranking system that considers, the editor rating, productivity, and peer-to-peer invigilation. This gives you access to the post-editors in the crowd who meet the criteria set by you, such as content domain, star rating, and speed.
So, what happens when you use a self-regulating crowd?
No more heavy HR work. You define a process, and it scales with the crowd.
No more QA project management – tasks are allocated, and it scales with the crowd.
KantanStream delivers today a fast, effective, and scalable crowd sourcing platform. It accelerates translation delivery and unlocks the potential to translate more content, helping develop better products and smoother user journeys for all.
We believe that KantanStream will change the way we think about, PEMT process management.