Count common words in two strings

Question

I have two strings:

a <- "Roy lives in Japan and travels to Africa"
b <- "Roy travels Africa with this wife"

I am looking to get a count of common words between these strings.

The answer should be 3.

• "Roy"

• "travels"

• "Africa"

being the common words

This is what I tried:

stra <- as.data.frame(t(read.table(textConnection(a), sep = " ")))
strb <- as.data.frame(t(read.table(textConnection(b), sep = " ")))

Taking unique to avoid repeat counting

stra_unique <-as.data.frame(unique(stra$V1))
strb_unique <- as.data.frame(unique(strb$V1))
colnames(stra_unique) <- c("V1")
colnames(strb_unique) <- c("V1")

common_words <-length(merge(stra_unique,strb_unique, by = "V1")$V1)

I need to this for a data set with over 2000 and 1200 strings. Total times I have to evaluate the string is 2000 X 1200. Any quick way, without using loops.

Answer 1

You can use strsplit and intersect from the base library:

> a <- "Roy lives in Japan and travels to Africa"
> b <- "Roy travels Africa with this wife"
> a_split <- unlist(strsplit(a, sep=" "))
> b_split <- unlist(strsplit(b, sep=" "))
> length(intersect(a_split, b_split))
[1] 3

Answer 2

Perhaps, using intersect and str_extract For multiple strings, you can either put them as a list or as vector

 vec1 <- c(a,b)
 Reduce(`intersect`,str_extract_all(vec1, "\\w+"))
 #[1] "Roy"     "travels" "Africa" 

For faster options, consider stringi

 library(stringi)
 Reduce(`intersect`,stri_extract_all_regex(vec1,"\\w+"))
 #[1] "Roy"     "travels" "Africa" 

For counting:

 length(Reduce(`intersect`,stri_extract_all_regex(vec1,"\\w+")))
 #[1] 3

Or using base R

  Reduce(`intersect`,regmatches(vec1,gregexpr("\\w+", vec1)))
  #[1] "Roy"     "travels" "Africa"