Bibliometric Analysis of Latent Dirichlet Allocation

Latent Dirichlet Allocation (LDA) has emerged as an important algorithm in big data analysis that finds the group of topics in the text data. It posits that each text document consists of a group of topics, and each topic is a mixture of words related to it. With the emergence of a plethora of text data, the LDA has become a popular algorithm for topic modeling among researchers from different domains. Therefore, it is essential to understand the trends of LDA researches. Bibliometric techniques are established methods to study the research progress of a topic. In this study, bibliographic data of 18715 publications that have cited the I,DA were extracted from the Scopus database. The software R and Vosviewer were used to carry out the analysis. The analysis revealed that research interest in LDA had grown exponentially. The results showed that most authors preferred Book Series followed by Conference Proceedings as the publication venue. The majority of the institutions and authors were from the USA, followed by China. The co-occurrence analysis of keywords indicated that text mining and machine learning were dominant topics in LDA research with significant interest in social media. This study attempts to provide a comprehensive analysis and intellectual structure of LDA compared to previous studies.

Automated summary

Latent Dirichlet Allocation (LDA) is an important algorithm in big data analysis that finds the group of topics in text data. Research interest in LDA has grown exponentially, with most authors and institutions from the USA and China. Text mining and machine learning are dominant topics in LDA research, with significant interest in social media.