| Marc Kupietz | 827a3c1 | 2019-09-18 22:09:33 +0200 | [diff] [blame] | 1 | #!/usr/bin/env Rscript |
| 2 | # |
| Marc Kupietz | 1242a5e | 2019-10-05 18:28:01 +0200 | [diff] [blame] | 3 | # Plot proportions of alternative expressions or spellings variants over time |
| Marc Kupietz | 827a3c1 | 2019-09-18 22:09:33 +0200 | [diff] [blame] | 4 | # |
| 5 | library(RKorAPClient) |
| 6 | library(ggplot2) |
| Marc Kupietz | 827a3c1 | 2019-09-18 22:09:33 +0200 | [diff] [blame] | 7 | library(plotly) |
| 8 | library(htmlwidgets) |
| 9 | |
| 10 | alternativesOverTime <- function(alternatives, years, kco = new("KorAPConnection", verbose=TRUE)) { |
| Marc Kupietz | 1242a5e | 2019-10-05 18:28:01 +0200 | [diff] [blame] | 11 | df <- expand_grid(Variant = alternatives, year = years) %>% |
| 12 | cbind(corpusQuery(kco, .$Variant, sprintf("textType = /Zeit.*/ & pubDate in %d", .$year))) %>% |
| 13 | group_by(year) %>% mutate(tokens = sum(totalResults)) %>% |
| Marc Kupietz | 865760f | 2019-10-07 19:29:44 +0200 | [diff] [blame^] | 14 | ci() %>% |
| 15 | rename(share=f) |
| 16 | g <- ggplot(data = df, mapping = aes(x = year, y = share, colour = Variant, fill = Variant)) + |
| Marc Kupietz | 0de631d | 2019-10-07 10:10:18 +0200 | [diff] [blame] | 17 | geom_freq_by_year_ci() + |
| Marc Kupietz | 827a3c1 | 2019-09-18 22:09:33 +0200 | [diff] [blame] | 18 | ggtitle(paste0(alternatives, collapse = " vs. ")) + |
| 19 | xlab("TIME") + |
| Marc Kupietz | 0de631d | 2019-10-07 10:10:18 +0200 | [diff] [blame] | 20 | ylab(sprintf("Observed frequency ratio")) |
| Marc Kupietz | 865760f | 2019-10-07 19:29:44 +0200 | [diff] [blame^] | 21 | ppp <- RKorAPClient::ggplotly(g) |
| Marc Kupietz | 827a3c1 | 2019-09-18 22:09:33 +0200 | [diff] [blame] | 22 | print(ppp) |
| 23 | df |
| 24 | } |
| 25 | |
| 26 | df <- alternativesOverTime(c('so "genannte.?"', '"sogenannte.?"'), (1995:2018)) |