The use of chronological clustering, that is, a particular contiguity constrained hierarchical clustering with stopping rules, has led to detect discontinuities in the corpus and thus determine time periods. Varying the α value leads to different sequencing, some finer than others.
Note that an α=0 is equivalent to performing HCCC without stopping rule, considering that all clusters can be considered homogeneous.
In this work, clustering method starts from the coordinates of the documents on the first CA axes. This joint use of CA and clustering allows us to simultaneously interpret CA and clustering results. This is fundamental because the finest aspects of the corpus structure can be hidden in the clustering results.
Xplortext R package (Bécue-Bertaut et al., 2022), available on R CRAN, has been used for the analysis. Xplortext web offers additional information on this package, as well as examples covered in detail.