Figure 1
From: Feature-rich multiplex lexical networks reveal mental strategies of early language learning
(A–C) Combining multiplex topology (A) and vector spaces (B) results in FERMULEX network (C); (D) kernel density estimates (KDEs) and ridgeline plots highlight conformity distribution for the frequency, length, and polysemy features in toddlers’ mental lexicon and the randomised variants; (E) Above—two-dimensional scatter plot of conformity vector space, where each point is colored according to the cluster the point belongs to (K-means algorithm); Below—distribution of word features within each cluster, where a kernel language emerges, i.e. the cluster labeled as D; (F) content characterisation of the kernel compared to a competitor from a k-core decomposition.
