Logiccrowd: integrating logic programming and crowdsourcing for mobile and pervasive computing platforms

Submission note: A thesis submitted in total fulfilment of the requirements for the degree of Doctor of Philosophy to the School of Engineering & Mathematic Sciences, College of Science, Health and Engineering, La Trobe University, Bundoora.

Tasks requiring computations have been traditionally solved by machines alone. Programmers have to provide a formal problem description and an algorithm to the machine so that it can compute a solution. In order to increase the capabilities of machine-based computational systems, more sophisticated algorithms can be used. However, we cannot solely depend on machines for certain tasks and queries, especially those that need critical thinking, analytical skills, human judgment, and an understanding of physical world context. Processing such tasks using machines alone could result in poor answers. This might be partly due to inadequate information stored in the database or the complexity of the problem. In such cases, despite the use of sophisticated algorithms, the sole use of machines cannot guarantee the quality of task completion. To address this issue, human integration with machine processing has been argued to be an option. With multiple skills, particularly in creative thinking, visual processing, planning, and analysis, humans, when integrated with machine computation for solving complicated task-related problems, are expected to produce higher levels of accuracy and efficiency. The emergence of the crowdsourcing paradigm has brought a dramatic change in the landscape of solving the problems. Crowdsourcing is an approach that involves human intelligence to solve the problems that may be rather difficult for computers to solve alone. At the same time, there has been an increasing need for easy-to-use programming languages. The declarative programming paradigm assumes programs to be descriptive in nature, focusing on the what rather than the how. Declarative programming languages and the declarative paradigm in general have received an increasing amount of widespread attention in the last decade. Due to the advantages of a high-level declarative manner of programming and the growing interest in crowdsourcing applications, the declarative approach to the development of crowdsourcing systems has tremendous potential. In this thesis, the tasks are designed to be completed by the crowd through mobile technology. This design is based on the evidence that mobile devices have become the most accessible tool and allow immediate communication between people nearby and across the globe through Internet connections. Also, since smart mobile devices are enriched with a set of embedded sensors and context data, they can support working-while-mobile, location-based services and peer-to-peer (P2P) communication. In this light, it may be argued that the use of mobile technology benefits crowdsourcing platforms. The focus of this research is on the development of a framework to construct mobile crowdsourcing applications by integrating the logic programming paradigm with crowd-based human processing. LogicCrowd is introduced as an innovative approach which combines conventional logic-based machine computation and the power of the crowd. It is built on top of existing social networking infrastructure, online crowdsourcing market platforms and peer-to-peer networks. The contribution of this thesis is two-fold. First, we propose a systematic approach to developing mobile crowdsourcing applications via the proposed Logic- Crowd framework. By extending the capabilities of Prolog, this approach enables a logic program to involve human input and capabilities in problem-solving and query-answering, rather than working only with a closed set of facts and rules in a local database. The approach also provides a novel extended unification scheme within the logic-programming paradigm. By applying the declarative programming paradigm over peer-to-peer networks, a LogicCrowd program is able to evaluate queries over multiple hops in mobile ad-hoc networks. Second, findings from a study involving extensive testing with a prototype system contribute insights into the use of mobile LogicCrowd-sourcing for real world applications. In this study, we demonstrate the versatility and usefulness of the system by introducing prototype applications based on our framework. The findings illustrate how the approach facilitates mobile users to leverage combinations of human intelligence and knowledge-based reasoning in order to solve complex problems. Furthermore, we analyse the energy characteristics of Logic- Crowd programs to understand factors that influence energy consumption across resource-limited mobile crowdsourcing platforms. The results indicate how mobile users can reduce the energy consumption of the mobile device when dealing with LogicCrowd queries.