从 R 中的文本中提取正确的湖名

Question

I am trying to extract the names of Lakes from some text that I have in R. The lakes are proper (capitalized) but will require me extracting a few words on either side of the word "Lake".

I tried a few things but nothing is working quite the way I want it to... in some cases, a sentence or the article may begin with "Lake" so there is no text before it. In some cases, the proper name may be 3 words (Lake St. Clair or Red Hawk Lake).

Example code to work with:

text <- paste("Lake Erie is located on the border of the United States and Canada.",
          "It is located nearby to Lake St. Clair and Lake Michigan.",
          "All three lakes have a history of high levels of Phosphorus.",
          "One lake that has not yet been impacted is Lake Ontario.")

This was maybe the closest I got-- pulling from another stack overflow but it's still not working out.

context <- function(text){splittedText <-strsplit(text,'',T) print(splitted Text) data.frame(before = head(c('',splittedText),-1),words=splittedText,after=tail(c(splittedText,''),-1))}

info <- context(text)
print(subset(info, words == 'Lake')

I would like to get either: 1) the proper lakes names extracted ("Lake Erie", "Lake St. Clair", etc.) OR 2) a dataframe with the words before and after "Lake". Ideally the first but I'm flexible at this point.

before <- c("","nearby to", "Clair and","impacted is")
Lake <- c("Lake","Lake","Lake","Lake")
after <- c("Erie is","St. Clair", "Michigan ","Ontario ")
output <- data.frame(cbind(before,Lake,after)); print(output)

Thanks in advance for the help!

Answer 1

You need to define some rules to extract words based on the data you have. Here I get the first word after the word "Lake" .

stringr::str_extract_all(text, "Lake \\w+")[[1]]
#[1] "Lake Erie"     "Lake St"       "Lake Michigan" "Lake Ontario" 

Or similarly in base R

regmatches(text, gregexpr("Lake \\w+", text))[[1]]

For the given text this almost works except for "Lake St. Clair" where it misses "Clair" part. To handle this we could have defined another rule where in case there is a dot after the next word of "Lake" , we extract two words but this would fail for "Lake Michigan" and "Lake Ontario" since they have full-stop following the word.

Answer 2

With stringi , we can use

library(stringi)
stri_extract_all_regex(text, "Lake\\s+\\w+")[[1]]
#[1] "Lake Erie"     "Lake St"       "Lake Michigan" "Lake Ontario" 

---

Or using str_match_all

library(stringr)
str_match_all(text, "Lake\\s+\\w+")[[1]][,1]