Genre, Computation, and the Varieties of Twentieth-Century U.S. Fiction » Wilkens_fig5

Figure 5. Eleven clusters identified by density-based analysis using the full feature set, arranged from earliest average publication date (purples and blues) to latest (yellows and reds) and projected into two-dimensional component space. Cluster dates are calculated relative to the average for all listed clusters (1951.8), which is later than the full corpus mean publication date (1936.6) and than the mean publication date of volumes in the text-only output (1943.0, figure 4). Two dimensions are used here, rather than the three in figure 4, to maximize readability in each case.

Print Friendly