Resilient Big Data Monetization

Rossi Kamal, Choong Seon HongarXiv 2015 · arXiv

Introduction

In the contemporary digital landscape, service providers leverage "Big Data" to extract value from user behaviors. This process typically involves using measurement tools—such as GPS traces or app usage metrics—to infer latent user interests like environmental context or emotional states. However, traditional monetization strategies often lack robustness; a single tool failure or a shift in user behavior can destabilize the entire inference model.

This paper proposes a theoretical framework to address this lack of resilience. By framing Big Data monetization through the lens of graph theory, the authors introduce a model for "Resilient Fact-Finding." This approach ensures that the insights derived from data are not only influential but also synergistic and capable of withstanding the dynamic nature of user behavior.

Theoretical Framework: Monetization as a Graph Problem

The core innovation of this research is the mapping of business monetization goals to the mathematical "Connected Dominating Set" (CDS) problem. In this model, measurement tools are treated as "dominators," while user interests are the "dominated" nodes.

Mapping monetization concepts to graph theory terms.

To achieve resilience, the authors define two key variables:

  • $k$-dominance: Ensures each user interest is monitored by at least $k$ different measurement tools, providing redundancy.
  • $m$-connectivity: Establishes strong ties between the measurement tools themselves, ensuring the network remains functional even if individual nodes fail.

The transition from a basic monetization problem to a resilient one is visualized as a shift toward these $k$ and $m$ constraints.

Conceptual overview of the Big Data monetization problem and the proposed Plutus solution.

The Plutus Algorithm

The authors prove that finding an optimal resilient set is an NP-hard problem, particularly within Unit Disk Graphs (UDG). To provide a practical solution, they introduce Plutus, a five-round greedy approximation algorithm:

  1. Isolation: Identifies the most influential measurement tools to serve as primary dominators.
  2. Domination: Establishes basic coverage where every user interest is linked to at least one tool.
  3. Synergy ($k$-dominance): Increases redundancy by ensuring each interest is connected to $k$ tools.
  4. Divergence ($m$-connectivity): Builds connectivity between the measurement tools.
  5. Sustainability: Strengthens the overall network structure to maintain connectivity against node removal.

Key Insights and Results

The paper's primary contribution is the formalization of "Resilient Big Data Monetization." Key takeaways include:

  • Mathematical Rigor: The formal proof of NP-hardness justifies the use of greedy approximation methods like Plutus.
  • Tunable Robustness: The $k/m$ framework allows service providers to adjust the level of resilience based on their specific needs and resource constraints.
  • Taxonomy of Monetization: The authors categorize existing schemes (e.g., social reputation, auction-based) and highlight their structural vulnerabilities compared to the proposed CDS model.

Limitations and Future Directions

While the theoretical foundation is robust, the paper lacks empirical validation using real-world datasets. Future research is needed to quantify the computational and operational costs of implementing high $k$ or $m$ values in live systems. Additionally, applying these "Resilient Fact-Finding" mechanisms to specific domains—such as inferring public transit patterns or monitoring health metrics—remains an open area for exploration.

Summary updated

BibTeX
@misc{kamal2015resilient,
  author = {Rossi Kamal and Choong Seon Hong},
  title = {{Resilient Big Data Monetization}},
  year = {2015},
  eprint = {1509.04545},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/1509.04545},
}
Figure