Differentiating, describing, and visualizing scientific space: A novel approach to the analysis of published scientific abstracts

This paper will develop and demonstrate a novel method for analyzing scientific indexes called Latent Semantic Differentiation. Using two distinct datasets comprised of scientific abstracts, it will demonstrate the procedure's ability to identify the dominant themes, cluster the articles accordingly, visualize the results, and provide a qualitative description of each cluster. Combined, the analyses will highlight the utility of the procedure for scientific document indexing, structuring university departments, facilitating grant administration, and augmenting ongoing research on scientific citation. Because the procedure is extensible to any textual domain, there are numerous avenues for continued research both within the sciences and beyond.