publish-date-bigram-tfidf.R

# ---- publish-date-bigram-tfidf
plot_bigram_tfidf <- function(dbin, start, stend, slice_unit="year", nmin=5, ntop=10, ncol=3) {
bigram_tf_idf <- dbin %>%
  subset(select=c(bigram,publish_date)) %>%
  mutate(Date = as.Date(substr(publish_date,1,8), format="%Y%m%d")) %>%
  filter(Date>=start & Date<=stend) %>%
  mutate(time_slice = floor_date(Date, unit = slice_unit)) %>%
  mutate(time_slice = format(time_slice, "%Y-%m-%d")) %>%
  count(time_slice, bigram) %>%
  filter(between(n, nmin, 10000)) %>%
  bind_tf_idf(bigram, time_slice, n) %>%
  arrange(desc(tf_idf))
bigram_tf_idf.plot <- bigram_tf_idf %>%
  mutate(bigram = factor(bigram, levels = rev(unique(bigram)))) %>%
  arrange(time_slice, tf_idf) %>%
  group_by(time_slice) %>% 
  top_n(n = ntop) %>% 
  ungroup()
plot_title <- paste("Ntop", ntop, "bigram tf-idf score", start, "to", stend, "slby", slice_unit)
bigram_tf_idf.plot %>%
  ggplot(aes(x = bigram, y = tf_idf, fill = time_slice)) +
  geom_col(show.legend = FALSE) +
  labs(x = "bi-grm", y = "tf-idf") +
  facet_wrap(~time_slice, ncol = ncol, scales = "free") +
  theme(text = element_text(size = 10)) +
  ggtitle(plot_title) +
  coord_flip()
}