Confidentiality and Integrity in Crowdsourcing Systems

Crowdsourcing information systems enlist a crowd of users to collaboratively build wide variety of information artifacts. Over the past decade, a large number of crowdsourcing information systems have been developed on the Internet such as Wikipedia, Linux, Yahoo! Answers, Stack Exchange, and numerous online social networks (OSNs).

Crowdsourcing information systems can be classified in many different ways. One of the classifications can be the nature of collaboration: explicit or implicit. In explicit collaboration systems (e.g., Wikipedia or Linux), users collaborate explicitly to build information artifacts. On the other hand, implicit collaboration systems let users collaborate implicitly to solve a problem for the system owners.

The privacy and security concerns in a crowdsourcing information system are dependent on how the ownerships of the content created by the system are managed. A crowdsourcing information system can use different ownership models: (a) creator owned, (b) system owned, and (c) group owned. In creator- owned systems, the creator of a data object is responsible for setting the access control constraints for the data object she creates. Examples of crowdsourcing information systems that use this model are online social networks and blogs. In system-owned crowdsourcing systems, individual contributors do not have the privilege of setting the access control policies for the content they create as part of their activities.

Integrity management is the review or establishment of different mechanisms to ensure the long-term integrity of artifacts in crowdsourcing systems. Crowdsourcing systems deal with two main challenges to preserve the integrity of their artifacts: assigning different capabilities to users, and evaluating users and their contributions. Two extreme approaches have been introduced for maintaining integrity in crowdsourcing systems: Wikipedia style and Linux style.

In this chapter, we investigate the integrity of Wikipedia articles as an example of integrity management in crowdsourcing systems. We first challenge its integrity by performing some experiments. Afterwards, we analyze the dump datasets from its website to find the reasons behind high integrity for few articles and low integrity for majority of them.

The tremendous success of crowdsourcing systems has attracted a lot of researchers. Prior work have investigated wide array of issues related to crowdsourcing such as visualization tools (Vegas et al. Studying cooperation and conflict between authors with history flow visualizations. 2004), (Vegas et al. Talk before you type: coordination in wikipedia. 2007), motivations for participation (Hoisl et al. Social rewarding in wiki systems motivating the community. 2007), the effects of coordination and collaboration (Wilkinson and Huberman. Cooperation and quality in wikipedia. 2007), vandalism analysis and detection (Kittur. He says, she says: conflict and coordination in wikipedia. 2007), (Priedhorsky et al. Creating, destroying, and restoring value in wikipedia. 2007), (Potthast et al. Automatic vandalism detection in wikipedia. 2008), (Smets et al. Automatic vandalism detection in wikipedia: towards a machine learning approach. 2008), reputation systems (Adler and de Alfaro. A content-driven reputation system for the wikipedia. 2007), (Lim et al. Measuring qualities of articles contributed by online communities. 2006), (Zeng et al. Computing trust from revision history. 2006), quality assurance and automatic quality measurement (Adler and de Alfaro. A content-driven reputation system for the wikipedia. 2007), (Blumenstock. Size matters: word count as a measure of quality on wikipedia. 2008), (Cross. Puppy smoothies: Improving the reliability of open, collaborative wikis. 2006), (Dondio and Barrett. Computational trust in web content quality: a comparative evalutation on the wikipedia project. 2007), (Lim et al. Measuring qualities of articles contributed by online communities. 2006), (Stvilia et al. Assessing information quality of a community-based encyclopedia. 2005), (Zeng et al. Computing trust from revision history. 2006), (Cross. Puppy smoothies: Improving the reliability of open, collaborative wikis. 2006), (Dondio and Barrett. Computational trust in web content quality: a comparative evalutation on the wikipedia project. 2007).