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.