A subset of the Microsoft Open Academic Graph (OAG), a dataset consisting of a unification of the Microsoft Academic Graph (MAG) and ArnetMiner (AMiner) academic graphs each respectively containing 166,192,182 and 154,771,162 papers. From this dataset, a subset of 37,732,477 papers with available abstracts and citation information were selected. These abstracts were preprocessed using stop words and stemming to form a vocabulary of 1,333 unique words. Together this vocabulary and corpus of papers were used to form a sparse 1,333 × 37,732,477 term-document matrix with 1,295,114,641 nonzeros, wherein each column represents a paper as a tf-idf vector. The resulting matrix was used as the X in the real world experiments. The symmetric graph Laplacian matrix S was then formed from the citation graph. Each of the 966,206,008 nonzeros of the resulting 37,732,477 × 37,732,477 matrix represents a citation between two papers.