A Trust Framework for the Collection of Reliable Crowd-Sourced Data

Abstract

With increasing access to the Internet through a multitude of devices, it is relatively easy to collect data from the public through a process called crowd-sourcing. Unfortunately, this approach has two problems: enticing users to contribute data, and determining if users have provided accurate data. The problem of inaccurate responses is compounded when a group of malicious users intentionally skew survey results in order to benefit themselves in some way. The first issue can be addressed through incentives, but this increases the need to solve the second issue since it is likely that users would produce fake data in order to obtain more of the offered incentives. We present a simple trust framework for addressing the issue of data quality in surveys, and illustrate its use with a real-world example. In this model, users are given access to the platform only when trusted users on the platform vouch for them. Although not fool proof, it does increase the quality of the collected data. Access to the survey results is used to incentivize users. We compared our solution to a traditional internet survey; we found that our solution reduced the number of invalid submissions by 9.29%.