원문정보
초록
영어
In the practical application of crowdsourcing, some unreliable workers have emerged due to profit driven. Their results seriously reduce the quality and bring about the initiator’s judgment biases. In this paper, we creatively put forward a crowdsourcing fraud detection method based on psychological behavior analysis to find out the spammer according to the psychological difference between deception and reliable behavior by means of Ebbinghaus forgetting curve. Furthermore, we constructed an online crowdsourcing experiment platform to verify the validity of our method. As a result, filtration recall and precision attained 0.93 and 0.79 respectively. Moreover, the overall LAM was decreased by 3.7% and AUC increased by 8% before and after filtration. Experimental results show that our method is effective and feasible.
목차
1. Introduction
2. Behavior Analysis based on Psychological Behavior Analysis
2.1. Ebbinghaus Forgetting Curve
2.2. Our Fraud Detection Algorithm
3. The Results and Analysis of Experiment
3.1. Construction and Evaluation Index of Experiment
3.2. Experimental Results and Analysis
4. Conclusion
Acknowledgements
References