Events

Online crowdsourcing platforms have enabled online social crowds to form group opinions. Existing crowd opinion aggregation methods were mainly developed for offline crowds and fail to consider the social influence commonly found in social crowds. Based on the social influence literature, we design a time-based decay function to account for social influence in social crowds. We propose a new opinion aggregation method, Social Crowd IQ (SCIQ), that considers social influence and prediction payoff for building a weighting model and uses all individuals rather than top contributors for opinion aggregation. Using two different datasets extracted from online crowdsourcing platforms, our experiments show that SCIQ outperforms baseline opinion aggregation methods. Additional functional testing shows that each of the three design elements in SCIQ is important in achieving the best crowd performance.