Get CBOW Matrix.
Get matrix with moving windows. Negative integer values indicate absence of a token at the respective position.
get_cbow_matrix(corpus, p_attribute, registry = Sys.getenv("CORPUS_REGISTRY"), matrix, window)
corpus |
a CWB corpus |
p_attribute |
a positional attribute |
registry |
the registry directory |
matrix |
a matrix |
window |
window size |
registry <- if (!check_pkg_registry_files()) use_tmp_registry() else get_pkg_registry() m <- get_region_matrix( corpus = "REUTERS", s_attribute = "places", strucs = 0L:5L, registry = registry ) windowsize <- 3L m2 <- get_cbow_matrix( corpus = "REUTERS", p_attribute = "word", registry = registry, matrix = m, window = windowsize ) colnames(m2) <- c(-windowsize:-1, "node", 1:windowsize)
Please choose more modern alternatives, such as Google Chrome or Mozilla Firefox.