improve comment parsing and add comment commentator redner matching
This commit is contained in:
+42
-4
@@ -24,12 +24,19 @@ collect_unique <- function(xs) xs %>% clear_na() %>% unique() %>% str_c(collapse
|
||||
# expects a tibble of redner and repairs
|
||||
repair_redner <- function(redner) {
|
||||
if (nrow(redner) == 0) return(redner)
|
||||
redner %>% mutate(fraktion = Vectorize(repair_fraktion)(fraktion)) %>% # fix fraktion
|
||||
group_by(id, vorname, nachname) %>%
|
||||
summarize(fraktion = collect_unique(fraktion),
|
||||
redner %>%
|
||||
filter(id != "10000") %>% # invalid id's
|
||||
mutate(fraktion = Vectorize(repair_fraktion)(fraktion)) %>% # fix fraktion
|
||||
group_by(id) %>%
|
||||
summarize(vorname = head(vorname, 1),
|
||||
nachname = head(nachname, 1),
|
||||
fraktion = collect_unique(fraktion),
|
||||
titel = longest_titel(titel),
|
||||
rolle_kurz = collect_unique(str_squish(rolle_kurz)),
|
||||
rolle_lang = collect_unique(str_squish(rolle_lang)))
|
||||
rolle_lang = collect_unique(str_squish(rolle_lang))) %>%
|
||||
ungroup() #%>%
|
||||
# arrange(id) %>%
|
||||
# distinct(vorname, nachname, fraktion, titel)
|
||||
}
|
||||
|
||||
repair_reden <- function(reden) {
|
||||
@@ -44,6 +51,37 @@ repair_talks <- function(talks) {
|
||||
talks
|
||||
}
|
||||
|
||||
# tries to find the correct redner id given a name
|
||||
# this is sufficient since every prename lastname combination in the bundestag is
|
||||
# unique (luckily :D)
|
||||
# returns a lookup table
|
||||
lookup_redner <- function(comments, redner) {
|
||||
tobereplaced <- "[-–—‑- ]"
|
||||
redner %>%
|
||||
unite(name, vorname, nachname, sep=".*") %>%
|
||||
mutate(name = str_replace_all(name, tobereplaced, ".*")) ->
|
||||
rs
|
||||
find_match <- function(komm) {
|
||||
if (komm == "") return (NA_character_)
|
||||
# I tried with agrep (levensthein distance) but results are better that way
|
||||
matches <- str_which(komm, rs$name)
|
||||
if (length(matches) == 0) return(NA_character_)
|
||||
rs[head(matches, 1), ]$id
|
||||
}
|
||||
comments %>%
|
||||
distinct(kommentator) %>%
|
||||
mutate(redner = Vectorize(find_match)(str_replace_all(kommentator, tobereplaced, "")))
|
||||
}
|
||||
|
||||
repair_comments <- function(comments, redner) {
|
||||
# try to find a redner id for each actual comment
|
||||
comments %>%
|
||||
filter(!is.na(kommentator)) %>%
|
||||
lookup_redner(redner) %>%
|
||||
left_join(comments, ., by="kommentator") %>%
|
||||
select(-kommentator)
|
||||
}
|
||||
|
||||
#' Repair parsed tables
|
||||
#'
|
||||
#' @export
|
||||
|
||||
Reference in New Issue
Block a user