TurKPF: TurKontrol as a Particle Filter
This work addresses computational bottlenecks for researchers using iterative crowdsourcing algorithms, but it is incremental as it primarily optimizes an existing method.
The authors tackled the computational inefficiency of TurKontrol, a POMDP-based algorithm for crowdsourced workflow control, by re-implementing it as TurKPF using a Particle Filter, resulting in nearly instantaneous action selection with similar performance to the original.
TurKontrol, and algorithm presented in (Dai et al. 2010), uses a POMDP to model and control an iterative workflow for crowdsourced work. Here, TurKontrol is re-implemented as "TurKPF," which uses a Particle Filter to reduce computation time & memory usage. Most importantly, in our experimental environment with default parameter settings, the action is chosen nearly instantaneously. Through a series of experiments we see that TurKPF and TurKontrol perform similarly.