R text analysis: Counting occurences of any combinations of words from two different keyword lists with a given distance of each other

user_5396970 0
Viewed 31

Thanks for reading. For a reserach project, I'm doing some text analysis. We are analyzing large texts (company reports) and I'm looking to count keyword frequencies within that text.

However, I have two lists of keywords, and I dont want to count the number of occurances of these words, but the number of times any two words from these lists appear within a certain distance from each other in the main text.

text <- c("The house is blue. The car is very big and red.")
words1 <- c("car", "house") 
words2 <- c("blue", "red")

The desired functionality should, for example, return 1 for distance 3. (Number of any combinations in given distance.)

I know about the text_count function from the stringb package and kwic from quantea. However, thats not really what Im looking for.

Thanks, any help is appreciated.

corpus quanteda nlp r
2 Answers

The quanteda package has the function fcm() that counts frequency of their co-occurrences.

require(quanteda)
txt <- c("The house is blue. The car is very big and red.")
toks <- tokens(txt) %>% tokens_tolower()
fcm(toks, window = 3, tri = FALSE)
#> Feature co-occurrence matrix of: 10 by 10 features.
#>         features
#> features the house is blue . car very big and red
#>    the     1     2  4    2 4   2    2   2   2   2
#>    house   2     0  2    1 2   1    1   1   1   1
#>    is      4     2  1    2 4   2    2   2   2   2
#>    blue    2     1  2    0 2   1    1   1   1   1
#>    .       4     2  4    2 1   2    2   2   2   2
#>    car     2     1  2    1 2   0    1   1   1   1
#>    very    2     1  2    1 2   1    0   1   1   1
#>    big     2     1  2    1 2   1    1   0   1   1
#>    and     2     1  2    1 2   1    1   1   0   1
#>    red     2     1  2    1 2   1    1   1   1   0
user_70500910

The following function returns true if the two words are in the text and at most at dist from one another.

text <- c("The house is blue. The car is very big and red.")
words1 <- c("car", "house") 
words2 <- c("blue", "red") 

fun <- function(x, text, dist) {
  txt <- strsplit(text, " ")[[1]]
  if(sum(x %in% txt) == length(x)) {
    d <- abs(diff(sapply(words1, \(x) grep(x, txt)))) - 1L
    unname(d <= dist)
  } else FALSE
}
fun(words1, text, 3)
#> [1] TRUE
fun(words2, text, 3)
#> [1] FALSE

Created on 2022-09-02 by the reprex package (v2.0.1)

user_82454060
Related