Data Science Overview

Principles

Definition

Data science is a multidisciplinary field that employs scientific methods, computational processes, and systems to extract knowledge and insights from structured and unstructured data. It investigates phenomena through the analysis and modeling of data to inform decision-making, solve complex problems, and advance understanding across diverse domains.

Ontological type

Core Methods

Data Sources

Ethical Issues

Key developments

Key Figures

Relational and Pattern Foundations era

Rakesh Agrawal [1] is associated with the University of Wisconsin–Madison [3] in this era. His key contributions include the mining of association rules in large databases, notably the papers "Mining association rules between sets of items in large databases" [6] (1993) and "Fast Algorithms for Mining Association Rules in Large Databases" [7] (1994), which established scalable methods for discovering patterns in large relational datasets and laid groundwork for subsequent data mining preprocessing and pattern discovery in relational models. Alex Pentland [2] has been affiliated with both the Massachusetts Institute of Technology [4] and Stanford University [5] in this era. His contribution is the Eigenfaces for Recognition [8] (1991), an early demonstration of PCA-based pattern recognition that helped establish data-driven approaches enabling computer vision and scalable analytics in the relational and pattern foundations.

Scalable Mining and Representations era

Li Fei-Fei [1] is a prominent researcher associated with Stanford University [3] and Princeton University [4] during the Scalable Mining and Representations era. Her key contribution, the ImageNet paper [5], introduced a large-scale hierarchical image database that enabled scalable pattern discovery and representation learning, becoming a foundational benchmark for vision research in this era. Li-Jia Li [2] is a prominent researcher associated with Stanford University [3] and Princeton University [4] during the Scalable Mining and Representations era. Her work on ImageNet [5] helped establish scalable datasets and hierarchical taxonomy, driving progress in data-driven representations and scalable mining in this era.

Deep Multimodal Self-Supervision era

Ilya Sutskever[1] is a prominent figure in the Deep Multimodal Self-Supervision era, with institutional affiliations at University of Toronto[3] and New York University[4]. His pivotal contributions include ImageNet Classification with Deep Convolutional Neural Networks[7], Sequence to Sequence Learning with Neural Networks[8], and Improving neural networks by preventing co-adaptation of feature detectors[9], which collectively established the viability of deep CNNs, sequence models, and regularization-driven data efficiency in this era. Ross Girshick[2] is a leading researcher in this era, with affiliations at University of Washington[5] and University of California, Berkeley[6]. His influential contributions include Rich Feature Hierarchies for Accurate Object Detection and Semantic Segmentation[10], Fast R-CNN[11], and Faster R-CNN: towards real-time object detection with region proposal networks[12], which advanced object-detection pipelines from staged approaches to real-time, scalable end-to-end systems in this era.