I have a dataset which looks like:

I want to create a data frame for a graph something like:

The B in the title is a different network from the above.

I tried to add columns but don't know what conditions to put to create such columns with values.

df %>% 
  mutate(from = ifelse(type == ____, ___, ____))

I want to create a network from the above something like

CodePudding user response：

library(tidyverse)

df <- tribble(~title, ~text, ~type, ~id,
"A",    "I like...",    NA,     "67340",
"C",    "I wish...",    "Question",     "c912",
"C",    "review1",  "Replies",  "a7869",
"C",    "review2",  "Replies",  "5769",
"C",    "review3",  "Answer",   "67340",
"C",    "Detecting...",     "Answer",   "28524",
"C",    "review1",  "Answer",   "80445",
"B",    "Let's..",  NA,     "4fcd",
"C",    "New...",   "Replies",  "5769")


df %>% mutate(conversation_group = na_if(title, "C")) %>%
  fill(conversation_group) %>%
  group_by(conversation_group) %>%
  transmute(from = id,
            to = replace_na(lag(id), first(conversation_group)),
            relation = case_when(is.na(type) ~ "Asked",
                                 type == "Question" ~ "Commented",
                                 type == "Replies" ~ "Replied",
                                 type == "Answer" ~ "Answered"),
            description = text) %>%
  ungroup() %>%
  select(-conversation_group)
#> # A tibble: 9 × 4
#>   from  to    relation  description 
#>   <chr> <chr> <chr>     <chr>       
#> 1 67340 A     Asked     I like...   
#> 2 c912  67340 Commented I wish...   
#> 3 a7869 c912  Replied   review1     
#> 4 5769  a7869 Replied   review2     
#> 5 67340 5769  Answered  review3     
#> 6 28524 67340 Answered  Detecting...
#> 7 80445 28524 Answered  review1     
#> 8 4fcd  B     Asked     Let's..     
#> 9 5769  4fcd  Replied   New...

^{Created on 2022-06-18 by the reprex package (v2.0.1)}