Using DocumentTermMatrix on a Vector of First and Last Names - r

Question

I have a column in my data frame (df) as follows:
> people = df$people
> people[1:3]
[1] "Christian Slater, Tara Reid, Stephen Dorff, Frank C. Turner"
[2] "Ice Cube, Nia Long, Aleisha Allen, Philip Bolden"
[3] "John Travolta, Uma Thurman, Vince Vaughn, Cedric the Entertainer"
The column has 4k+ unique first/last/nick names as a list of full names on each row as shown above. I would like to create a DocumentTermMatrix for this column where full name matches are found and only the names that occur the most are used as columns. I have tried the following code:
> people_list = strsplit(people, ", ")
> corp = Corpus(VectorSource(people_list))
> dtm = DocumentTermMatrix(corp, people_dict)
where people_dict is a list of the most commonly occurring people (~150 full names of people) from people_list as follows:
> people_dict[1:3]
[[1]]
[1] "Christian Slater"
[[2]]
[1] "Tara Reid"
[[3]]
[1] "Stephen Dorff"
However, the DocumentTermMatrix function seems to not be using the people_dict at all because I have way more columns than in my people_dict. Also, I think that the DocumentTermMatrix function is splitting each name string into multiple strings. For example, "Danny Devito" becomes a column for "Danny" and "Devito".
> inspect(actors_dtm[1:5,1:10])
<<DocumentTermMatrix (documents: 5, terms: 10)>>
Non-/sparse entries: 0/50
Sparsity : 100%
Maximal term length: 9
Weighting : term frequency (tf)
Terms
Docs 'g. 'jojo' 'ole' 'piolin' 'rampage' 'spank' 'stevvi' a.d. a.j. aaliyah
1 0 0 0 0 0 0 0 0 0 0
2 0 0 0 0 0 0 0 0 0 0
3 0 0 0 0 0 0 0 0 0 0
4 0 0 0 0 0 0 0 0 0 0
5 0 0 0 0 0 0 0 0 0 0
I have read through all the TM documentation that I can find, and I have spent hours searching on stackoverflow for a solution. Please help!

Answer 1

The default tokenizer splits text into individual words. You need to provide a custom function
commasplit_tokenizer <- function(x)
unlist(strsplit(as.character(x), ", "))
Note that you do not separate the actors before creating the corpus.
people <- character(3)
people[1] <- "Christian Slater, Tara Reid, Stephen Dorff, Frank C. Turner"
people[2] <- "Ice Cube, Nia Long, Aleisha Allen, Philip Bolden"
people[3] <- "John Travolta, Uma Thurman, Vince Vaughn, Cedric the Entertainer"
people_dict <- c("Stephen Dorff", "Nia Long", "Uma Thurman")
The control options didn't work with just Coprus, I used VCorpus
corp = VCorpus(VectorSource(people))
dtm = DocumentTermMatrix(corp, control = list(tokenize =
commasplit_tokenizer, dictionary = people_dict, tolower = FALSE))
All of the options are passed within control, including:
tokenize - function
dictionary
tolower = FALSE
Results:
as.matrix(dtm)
Terms
Docs Nia LOng Stephen Dorff Uma Thurman
1 0 1 0
2 0 0 0
3 0 0 1
I hope this helps