Game Development Reference
In-Depth Information
Fig. 2.2 Example of Duolingo's interface for translating real Web. The application aid the student
with translation of individual work, making him practice with vocabulary and able to translate even
complex sentences
But Duolingo has also its crowdsourcing part: as a practice, its students can
translate real sentences from web pages written in the foreign language to English.
For this, a convenient interface (see Fig. 2.2 ) is provided. The student may also review
and rate other translations. For these activities, he is also rewarded with achievement
points. Using a redundancy principle and a wisdom of the crowd, Duolingo is able
to output very accurate translations of a real web pages using only a group of lay
translators that are actually just learning the language.
2.5.6 Crowdsourcing Games
The gamification aims to solve the problem of limited motivation to participate in
a crowdsourcing by using game-like elements. As such, it may be perceived as
a springboard towards “full” crowdsourcing games —a part of the crowdsourcing
approach family. These games emerged as an alternative to solving computational
problems, hard or impossible to be solved by machine computation (which includes
acquiring semantic structures), via aggregation of knowledge provided by many
non-expert users (e.g., for image annotation) [ 62 ]. Crowdsourcing games transform
problems into games that motivate players to solve them via fun and thus eliminate
the need to pay them. As many game instances can be played simultaneously, they
are suitable for larger scale problems divisible into smaller tasks. Compared to other
crowdsourcing techniques, the knowledge gained in crowdsourcing games is not just
a by-product of another user activity (e.g., annotating web resources for personal
use), but the primary objective, so their design is tuned to maximize that ability. The
crowdsourcing games are discussed in detail in the Chap. 3 .
2.6 Discussion
To sum up, there is a variety of approaches for building web semantics rang-
ing from manual through crowdsourcing to automated ones. Semantics discovery
approaches are evaluated with respect to quantity (number of instances retrieved,
 
Search WWH ::




Custom Search