Overlapping Community Detection in Statistical Keyword Co-Occurrence Network by Mixed-SCORE

Keyword co-occurrence networks, constructed by the co-occurrence relationships of keywords in academic publications, have become essential tools for exploring and understanding research topics within specific academic fields. However, traditional community detection methods often assume disjoint partitions and fail to capture the inter-topic nature of research fields. To the best of our knowledge, we are the first to analyze the research topics in statistics by constructing a keyword co-occurrence network using papers published in 42 statistical journals from 1992 to 2021. To correct for degree heterogeneity, we apply the Mixed-SCORE method on the 40-core network, resulting in ten overlapping communities, such as the time series analysis, high-dimensional analysis, and hypothesis testing communities. For the first time, the presence of overlapping communities within the keyword co-occurrence network is uncovered, indicating the interconnections between different research topics. In addition, to evaluate the effectiveness of the Mixed-SCORE method, we compare the results with other overlapping community detection methods and demonstrate its good performance. Our study contributes to the understanding of research topics and interconnections between multiple topics in statistics. The innovative use of overlapping community detection on keyword co-occurrence networks provides a new perspective for analyzing research patterns in various disciplines.