Topic modeling in density functional theory on citations of condensed matter electronic structure packages

With an increasing number of new scientific papers being released, it becomes harder for researchers to be aware of recent articles in their field of study. Accurately classifying papers is a first step in the direction of personalized catering and easy access to research of interest. The field of Density Functional Theory (DFT) in particular is a good example of a methodology used in very different studies, and interconnected disciplines, which has a very strong community publishing many research articles. We devise a new unsupervised method for classifying publications, based on topic modeling, and use a DFT-related selection of documents as a use case. We first create topics from word analysis and clustering of the abstracts from the publications, then attribute each publication/paper to a topic based on word similarity. We then make interesting observations by analyzing connections between the topics and publishers, journals, country or year of publication. The proposed approach is general, and can be applied to analyze publication and citation trends in other areas of study, beyond the field of Density Function Theory.

Automated summary

With the increasing number of scientific papers, it becomes difficult for researchers to keep up with recent articles in their field. In this study, a new unsupervised method using topic modeling is proposed to classify publications, with a focus on Density Functional Theory (DFT). By analyzing word similarity, the method attributes each publication to a specific topic and provides interesting observations on connections between topics and publishers, journals, country or year of publication. This approach is general and can be applied to analyze publication and citation trends in other areas of study beyond DFT.