Best way to clean free text then turn into a transaction dataset

I have survey information that contains free text that I would like to clean then put into a transaction dataset to run in the arules R package. Right now the text looks like this.

id | Answers    
1  | John thinks that the product is not worth the price
2  | Amy believes that the functionality is well above expectations

Here's what I'm trying to do:

1 | John | thinks   | Product       | Not   | Worth | Price    
1 | Amy  | Believes | Functionality | Above | Expectations

Right now I have been able to clean the data using tm package but I don't know what is the best way to convert it to a transaction dataset. I've turned the information into all lowercase and removed the stop words.

Let's just say my data is in data frame called "Questions". I am unable to convert the corpus into a transaction dataset after I have cleaned it.

Solution

You can try:

library(stringr)
str_split(data$Answers, " ")

The output is a list:

[[1]]
 [1] "John"    "thinks"  "that"    "the"     "product" "is"      "not"     "worth"   "the"     "price"  

[[2]]
[1] "Amy"           "believes"      "that"          "the"           "functionality" "is"           
[7] "well"          "above"         "expectations"

Edit:

Removing duplicates using the unique function:

my_list <- str_split(data$Answers, " ")
lapply(my_list , unique)

[[1]]
[1] "John"    "thinks"  "that"    "the"     "product" "is"      "not"     "worth"   "price"  

[[2]]
[1] "Amy"           "believes"      "that"          "the"           "functionality" "is"           
[7] "well"          "above"         "expectations"