Big data clustering techniques based on Spark: a literature review.

Area of Science:

• Data Mining
• Machine Learning
• Pattern Recognition

Background:

• Clustering is a key unsupervised learning technique challenged by massive data growth.
• Traditional methods struggle with Big Data, necessitating Big Data platform integration.
• Apache Spark offers fast, distributed processing for Big Data challenges.

Purpose of the Study:

• To systematically survey existing Apache Spark-based clustering methods.
• To evaluate these methods against Big Data characteristics.
• To propose a novel taxonomy for Spark-based clustering.

Main Methods:

• Systematic literature review of Spark-based clustering studies (2010-2020).
• Analysis of clustering methods concerning Big Data properties.
• Development of a new classification taxonomy.

Main Results:

• Identified and categorized existing Spark-based clustering approaches.
• Assessed the suitability of current methods for Big Data.
• Established a comprehensive overview of the field.

Conclusions:

• Spark-based clustering is an emerging research area with significant potential.
• A structured taxonomy is needed to understand and advance the field.
• Further research is required to address Big Data clustering challenges effectively.