Accountability management of human-based activities in collaborative computing

Abstract

Providing accountability in a collaborative computer system based on human-based activity is a key challenge. The identification and definition of responsible users to ensure proper and correct data transfer demand the design of a system with a set of rules that limit misconduct and verify the correctness of actions. As individual accountability is pivotal in collaborative computing, a trustworthy system that prevents misconduct is crucial. In many distributed systems, trust is established through dependence on third parties, which requires the distribution of sensitive data and results in loss of data sovereignty. New concepts are therefore required that are not relying on trusted third parties, but ensure that misconduct in such an environment is recognized and traced back to the responsible user. With the introduction of the very first Distributed Ledger Technology (DLT) in 2008 named Bitcoin, it became possible for the first time to establish trust between peers in a peer-to-peer (P2P) network without relying on a trusted third party. Throughout this thesis project, an accountability model was implemented, ensuring that activities carried out by people are verifiable and trustworthy using a decentralized system along with the engagement of a participating collective. The generation of reliable and verifiable provenance information, based on the PROV standard, plays a key role in this implementation. Moreover, a cryptographically enforced agreement, such as a smart contract in Ethereum is utilized to achieve the ability to track the transaction chain of the involved users and thus their actions. It emerged to a reliable candidate solution for detecting misconduct and verification of the correctness of their actions. For demonstration and evaluation purposes a functional prototype was implemented that extends the SmartSociety programming framework by a working scenario.