Abstract

AI-driven services and data collecting-based applications are these days the talk of the town in the field of computer science and beyond. This is often obvious since one can effortlessly take note, by looking around, how this field has ended up fundamental in our day to day lives beginning from business intelligence and market analysis down to virtual personal assistants (e.g., Siri, Google Now) and social media analysis (e.g., people you may know on Facebook). However, the research communities anticipate a turn down in the revolution of AI-driven services and data collecting-based applications due to the deficiency within the accessibility of huge data that ought to be collected or (pre-)trained using machine learning technologies. Mainly, collecting and integrating the big complementary data scattered across foundations and countries entails high costs and management challenges associated with finding and getting on board multiple data providers. In this thesis, we tackle this problem by designing a data market platform on top of the cloud computing technology. The data market platform helps the data providers and data consumers find and meet each other, while the cloud technology provides the required computing resources to execute computational tasks associated with data processing. This thesis starts by searching and investigating the most efficient business theories to model the data market platform. As a result of our search, the two-sided market theory has been proposed as a successful underlying model to design the data market platform.

The two-sided market theory has been improved in this thesis to handle challenges associated with considering data as an economic good on the one hand, and reshaping the business of the cloud computing to act as a data market platform on the other hand. Mainly, we introduce a novel game theoretical model (Two-sided game), which consists of a mix of cooperative and competitive strategies. The players of the game are the big data providers, cloud computing platform, and data consumers. The strategies of the players are modeled using the two-sided market theory that takes into consideration the network effects (externalities) among involved parties. The externalities refer to the mutual impact of the number of data providers and data consumers on each other. The objective of this game is to enable the cloud to be an active platform that can help big data service providers reach a wider set of customers and cloud users (i.e., data consumers) to be exposed to a larger and richer variety of data to run their data analytic tasks. The proposed game has been improved further to deliver complementary data services among multiple data providers over a cloud intermediary platform. More specifically, we formalize the problem as an extended two-sided market model by courting on one side some influential data providers in order to attract other data providers on the same side to form a bundling of data services. The final game aims to dynamically distribute the cloud computing resources among computational tasks of data providers to maximize the social welfare of all involved parties. The game has been supported by a mechanism to handle potential undesired behaviours such as the greedy and irrationality behaviours of involved parties. The game also provides a clear pricing mechanism that estimates the monetary value of data considering the actual need of the data consumers.

The thesis ends up by involving the blockchain technology in the process of monetizing data. The blockchain technology has recently proved to be an efficient solution for guaranteeing the security of data transactions in data trading scenarios. The benefits of the blockchain in this domain have been shown to span over several crucial security and privacy aspects such as verifying the identities of data providers, detecting and preventing malicious data consumers, and regulating the trust relationships between the data trading parties. However, the cost and economic aspects of using this solution such as the pricing of the mining process have not been addressed yet. In fact, using the blockchain entails high operational costs and puts both the data providers and miners in a continuous dilemma between delivering high-quality security services and adding supplementary costs. In addition, the mining leader requires an efficient mechanism to select the tasks from the mining pool and determine the needed computational resources for each particular task in order to maximize its payoff. Motivated by these two points, we propose in this thesis a novel game theoretical model based on the two-sided market approach that helps both the data providers and miners determine the monetary reward and computational resources, respectively.