Core Concepts
This paper presents a comprehensive data-driven survey of the research landscape on decomposition-based evolutionary multi-objective optimization, particularly multi-objective evolutionary algorithm based on decomposition (MOEA/D), leveraging advanced data mining techniques to uncover prominent research topics, trends, collaborations, and citations.
Abstract
This paper conducts a comprehensive data-driven survey of the research landscape on decomposition-based evolutionary multi-objective optimization, particularly multi-objective evolutionary algorithm based on decomposition (MOEA/D). The authors constructed a heterogeneous knowledge graph encompassing over 5,400 papers, 10,000 authors, 400 venues, and 1,600 institutions to enable a large-scale analysis.
The key highlights of the survey include:
General Statistics:
The publication trend shows a robust quadratic growth, with the number of MOEA/D-related publications reaching nearly 800 in 2023.
The research community has expanded rapidly, with over 10,000 authors contributing to MOEA/D by 2024.
MOEA/D research is globally distributed, with China, the United States, the United Kingdom, India, and Spain as the top contributing regions.
Journals account for 76% of MOEA/D publications, led by Applied Soft Computing, IEEE Transactions on Evolutionary Computation, and Swarm and Evolutionary Computation.
Topic Modeling:
The authors employed advanced topic modeling techniques, leveraging BERT-based embeddings and hierarchical clustering, to identify 83 distinct research topics in the MOEA/D literature.
These topics were categorized into two main types: methodological enhancements and extensions (40 topics), and application-driven research (43 topics).
The methodological topics cover themes such as weight vector settings, archives, estimation of distribution methods, penalties, performance indicators, parallelization, robustness, and visualization.
The application-driven topics span diverse domains, including scheduling problems, routing problems, computer science applications (network science, software engineering, computer vision, machine learning), and engineering design.
Citation and Collaboration Network Analysis:
The citation network analysis revealed the disruptiveness and evolution of MOEA/D research, identifying landmark studies that have significantly influenced the field.
The collaboration network analysis uncovered the most active researchers in the MOEA/D community and their patterns of collaboration.
The comprehensive data-driven approach employed in this survey provides a holistic view of the MOEA/D research landscape, offering valuable insights for both researchers and practitioners in the field of evolutionary multi-objective optimization.
Stats
"The publication trend shows a robust quadratic growth, with the number of MOEA/D-related publications reaching nearly 800 in 2023."
"The research community has expanded rapidly, with over 10,000 authors contributing to MOEA/D by 2024."
"Journals account for 76% of MOEA/D publications, led by Applied Soft Computing, IEEE Transactions on Evolutionary Computation, and Swarm and Evolutionary Computation."
Quotes
"This paper presents a comprehensive data-driven survey of the research landscape on decomposition-based evolutionary multi-objective optimization, particularly multi-objective evolutionary algorithm based on decomposition (MOEA/D)."
"The authors constructed a heterogeneous knowledge graph encompassing over 5,400 papers, 10,000 authors, 400 venues, and 1,600 institutions to enable a large-scale analysis."
"The comprehensive data-driven approach employed in this survey provides a holistic view of the MOEA/D research landscape, offering valuable insights for both researchers and practitioners in the field of evolutionary multi-objective optimization."