CODEWITHSUNDEEP
rstring
Debbie
Debbie

Reputation: 401

Matching similar string vectors and return non-matching element

I have 2 datasets containing similar string vectors (product titles). The only difference between the strings in both datasets is absence/presence of special characters.

Now, my problem is to match the corresponding string vectors and return the non-matching element(s) (which should be special character(s) in each case). There can be many non-matching special characters in a single string.

For e.g. I have 2 texts:

Text 1: Analog Science Fiction and Fact February 1995
Text 2: Analog Science Fiction and Fact, February 1995

Is there an R function to return the non-matching element(s) only?

This is how I approached the problem

S.vector <- strsplit(Acceptdata['Text.1'][1,],' ')
S.vector
# [[1]]
# [1] "Analog"   "Science"  "Fiction"  "and"      "Fact"     "February" "1995"    

F.vector <- strsplit(Acceptdata['Text.2'][1,],' ')
F.vector
# [[1]]
# [1] "Analog"   "Science"  "Fiction"  "and"      "Fact,"    "February" "1995"

l.S.vector <- tolower(S.vector)
l.F.vector <- tolower(F.vector)
grep("l.S.vector",l.F.vector,invert=T,value=T)
# [1] "c(\"analog\", \"science\", \"fiction\", \"and\", \"fact,\", \"february\", \"1995\")"

Any help is greatly appreciated.

When I'm trying to run the algorithm for the entire dataset(~500 vectors) its throwing an error as is.character(a) is not TRUE.

The procedure I followed: 

common <- function(a,b) { 
  for (i in seq_along(a)) 
    for (j in seq_along(b)) 
    i2 <- strsplit(tolower(i),'') 
    j2 <- strsplit(tolower(j),'') 
    if(length(i2) < length(j2)) { 
      i2[(length(i2)+1):length(j2)] <- ' ' 
    } else if(length(i2) > length(j2)) { 
      b2[(length(b2)+1):length(a2)] <- ' ' 
    } 
    LCS(i2,j2) 
} 

z <- common(a,b) 
Error: is.character(a) is not TRUE

Any idea on where did I go wrong?

Upvotes: 1

Views: 687

Answers (1)

Thomas
Thomas

Reputation: 44565

I'm totally clear on your intended output, but I think this will help you get there. It uses the LCS function from the qualV package.

library("qualV")
common <- function(a,b) {
    a2 <- strsplit(a,'')[[1]]
    b2 <- strsplit(b,'')[[1]]
    if(length(a2) < length(b2)) {
        a2[(length(a2)+1):length(b2)] <- ' '
    } else if(length(a2) > length(b2)) {
        b2[(length(b2)+1):length(a2)] <- ' '
    }
    LCS(a2,b2)
}

Here's an example using your two strings:

a <- 'Analog Science Fiction and Fact February 1995'
b <- 'Analog Science Fiction and Fact, February 1995'
z <- common(a,b)
paste0(z$LCS, collapse = '') # common string
# [1] "Analog Science Fiction and Fact February 1995"
z$b[which(!seq(1,max(z$vb)) %in% z$vb)] # non-matching elements in `b`
# [1] ","
z$a[which(!seq(1,max(z$va)) %in% z$va)] # non-matching elements in `a`
# character(0)

Here's an example using two strings that have more differences:

a <- 'Analog! SCIENCE Fiction and Fact Feb. 1995'
b <- 'Analog Science Fiction & Fact (February 1995)'
z <- common(a,b)
paste0(z$LCS, collapse = '') # common string
# [1] "Analog S Fiction  Fact Feb 1995"
z$b[which(!seq(1,max(z$vb)) %in% z$vb)] # non-matching elements in `b`
# [1] "c" "i" "e" "n" "c" "e" "&" "(" "r" "u" "a" "r" "y"
z$a[which(!seq(1,max(z$va)) %in% z$va)] # non-matching elements in `a`
# [1] "!" "C" "I" "E" "N" "C" "E" "a" "n" "d" "."

Upvotes: 1

Related Questions