Demand-side energy management reimagined: A comprehensive literature analysis leveraging large language models

Abstract

The landscape of Demand-Side Energy Management (DSM) research is rapidly evolving, shaped by technological innovations and policy developments. This paper presents an exhaustive bibliometric analysis and methodological framework to explore the research trends within the DSM domain. By synthesizing data from Scopus and OpenAlex, we compile a comprehensive dataset of DSM publications that serve as the basis for our analysis. Through rigorous data acquisition and cleaning, we ensure the reliability and relevance of our dataset. We employ state-of-the-art Large Language Models (LLMs) and topic modeling techniques, including GPT and BERTopic, to perform semantic analysis and uncover thematic structures within the literature. Statistical analysis of the literature dataset reveals a steady increase in DSM publications, with significant contributions from prestigious journals and institutions worldwide. We observe that articles are the predominant publication type, while reviews often cite more references and receive higher citation counts. The distribution of publications over time indicates a growing interest in DSM, particularly since 2014. Geographical mapping of institutions highlights key regions contributing to DSM research, with notable outputs from Europe, North America, and East Asia. Coupled with citation network analysis, our approach reveals the influential works and emerging trends that define the scientific progression of DSM research. Our unsupervised topic modeling, powered by BERTopic, clusters the publications into distinct themes, while our advanced visualization techniques using UMAP and t-SNE provide insights into the semantic space of DSM literature. The resulting thematic classification is presented in a hierarchical structure, offering a comprehensive understanding of the field’s focus areas. Our citation network analysis, utilizing force-directed graph computation and edge-bundling algorithms, maps the interconnectivity and impact of research contributions, providing a dynamic view of the field’s evolution. This study not only charts the landscape of DSM research but also offers a methodological blueprint for future bibliometric analyses. The insights gained from this multi-faceted exploration serve as a valuable resource for researchers, policymakers, and industry practitioners looking to navigate the complexities of DSM and contribute to its scientific advancement.