The Research on the Application of Qualitative Mapping in MapReduce

Jiali Feng, Guanglin Xu, Xiaolin XuAnnals 2015 · DOI

Abstract

This paper explores a theoretical bridge between Attribute Theory and Big Data processing by proposing "Qualitative Mapping" as a formal mathematical framework. This framework enhances data transformation and categorization within the MapReduce ecosystem, allowing for the systematic conversion of raw quantitative data into meaningful conceptual categories.

Introduction

While MapReduce is a powerhouse for sorting and grouping large-scale data, its role in transforming raw numbers into qualitative concepts is often handled in an ad-hoc manner. In Big Data environments, where data is frequently semi-structured or unstructured, defining the relationship between a value (quantity) and its meaning (quality) is a significant challenge. This research addresses how to mathematically formalize the process of turning raw data points into qualitative judgments across distributed systems.

Methodology

The authors utilize Attribute Theory to create a formal model for data transformation. Central to this is the "Simplest Qualitative Mapping," a mathematical function $\tau$ that maps a quantity $x$ to a quality $p(u)$ based on a defined criterion range $[\alpha, \beta]$. A novel Feature Extraction Operator ($\in_?$) serves as the engine for the Map phase, determining if a value satisfies a qualitative criterion.

For complex objects requiring multi-attribute analysis, the model extends into $n$-dimensional space, where qualitative criteria form a hypercube or grid. In the MapReduce integration:

  1. Map Phase: Raw data pairs are transformed into quality/value pairs.
  2. Shuffle/Reduce Phase: Data is grouped by quality, allowing the system to determine the boundaries of categories across the entire dataset.
Figure 3: MapReduce workflow for categorizing student scores into grades.

Applications and Insights

The paper illustrates these concepts through two primary examples:

  • Education: Mapping numerical student scores to letter grades to demonstrate how MapReduce identifies score ranges for specific categories.
  • Medical Diagnostics: Categorizing patients based on multiple factors such as Fasting Blood Glucose (FBG), Cholesterol (TC), Triglycerides (TG), and HDL-C levels.
Figure 5: Multi-attribute qualitative mapping representing a diagnostic region in a 3D space of TC, TG, and HDL-C.

The core insight is the formalization of the "quantity-to-conception" leap. By treating this as a mathematical transformation, it can be optimized for parallel computing. In multi-attribute scenarios, "quality" is represented as a region in a multi-dimensional grid, which MapReduce can process in parallel.

Conclusion

The research successfully establishes a formal mathematical framework for qualitative mapping using set theory. It demonstrates that the standard MapReduce workflow is functionally equivalent to solving for qualitative criteria in Attribute Theory. While the study is primarily theoretical and lacks empirical performance benchmarks, it provides a foundation for applying qualitative mapping to complex unstructured data, potentially bridging the gap between raw data processing and high-level artificial intelligence.

Summary updated

BibTeX
@article{feng2015research,
  author = {Jiali Feng and Guanglin Xu and Xiaolin Xu},
  title = {{The Research on the Application of Qualitative Mapping in MapReduce}},
  journal = {Annals},
  year = {2015},
  doi = {10.1007/s40745-015-0054-8},
}
Figure