parsent

parsent is a collection of tools used to parse sentences. The package is a wrapper for the NLP/openNLP packages that simplifies and extends the user experience.

Table of Contents

• Function Usage
• Installation
• Contact
• Demonstration
  • Load the Packages/Data
  • Create Annotator
  • Parsing
  • Plotting
  • Get Subject, Verb, and Direct Object
    • Subject
    • Predicate Verb
    • Direct Object

Function Usage

Functions typically fall into the task category of (1) parsing, (2) converting, & (3) extracting. The main functions, task category, & descriptions are summarized in the table below:

Function	Task	Description
parser	parsing	Parse sentences into phrases
parse_annotator	parsing	Generate OpenNLP parser required by parser function
as_tree	converting	Convert parser output into tree form
as_square_brace	converting	Convert parser output in square brace form (vs. round)
as_square_brace_latex	converting	Convert parser output LaTeX ready form
get_phrases	extracting	Extract phrases from parser output
get_phrase_type	extracting	Extract phrases one step down the tree
get_phrase_type_regex	extracting	Extract phrases at any level in the tree (uses regex)
get_leaves	extracting	Extract the leaves (tokens or words) from a phrase
take	extracting	Select indexed elements from a vector

Installation

To download the development version of parsent:

Download the zip ball or tar ball, decompress and run R CMD INSTALL on it, or use the pacman package to install the development version:

if (!require("pacman")) install.packages("pacman")
pacman::p_load_gh(c(
    "trinker/textshape", 
    "trinker/coreNLPsetup",          
    "trinker/parsent"
))

Contact

You are welcome to:

• submit suggestions and bug-reports at: https://github.com/trinker/parsent/issues
• send a pull request on: https://github.com/trinker/parsent/
• compose a friendly e-mail to: tyler.rinker@gmail.com

Demonstration

Load the Packages/Data

if (!require("pacman")) install.packages("pacman")

pacman::p_load(parsent, magrittr)

txt <- c(
    "Really, I like chocolate because it is good. It smells great.",
    "Robots are rather evil and most are devoid of decency.",
    "He is my friend.",
    "Clifford the big red dog ate my lunch.",
    "Professor Johns can not teach",
    "",
    NA
)

Create Annotator

if(!exists('parse_ann')) {
    parse_ann <- parse_annotator()
}

Parsing

(x <- parser(txt, parse.annotator = parse_ann))

## [[1]]
## [1] "(TOP (S (S (ADVP (RB Really))(, ,) (NP (PRP I)) (VP (VBP like) (NP (NN chocolate)) (SBAR (IN because) (S (NP (PRP it)) (VP (VBZ is) (ADJP (JJ good)))))))(. .) (NP (PRP It)) (VP (VBZ smells) (ADJP (JJ great)))(. .)))"
## 
## [[2]]
## [1] "(TOP (S (S (NP (NNP Robots)) (VP (VBP are) (ADJP (RB rather) (JJ evil)))) (CC and) (S (NP (RBS most)) (VP (VBP are) (ADJP (JJ devoid) (PP (IN of) (NP (NN decency))))))(. .)))"
## 
## [[3]]
## [1] "(TOP (S (NP (PRP He)) (VP (VBZ is) (NP (PRP$ my) (NN friend)))(. .)))"
## 
## [[4]]
## [1] "(TOP (S (NP (NNP Clifford)) (NP (DT the) (JJ big) (JJ red) (NN dog)) (VP (VBD ate) (NP (PRP$ my) (NN lunch)))(. .)))"
## 
## [[5]]
## [1] "(TOP (S (S (NP (NNP Professor) (NNP Johns)) (VP (MD can))) (RB not) (VB teach)))"
## 
## [[6]]
## [1] NA
## 
## [[7]]
## [1] NA

Note that the user may choose to use CoreNLP as a backend by setting engine = "coreNLP". To ensure that coreNLP is setup properly use check_setup.

Plotting

par(mar = c(0,0,0,.7) + 0.2)
plot(x[[2]])

par(
    mfrow = c(3, 2),
    mar = c(0,0,1,1) + 0.1
)
invisible(lapply(x[1:5], plot))

Get Subject, Verb, and Direct Object

Subject

get_phrase_type(x, "NP") %>%
    take() %>%
    get_leaves()

## [[1]]
## [1] "I"
## 
## [[2]]
## [1] "Robots"
## 
## [[3]]
## [1] "He"
## 
## [[4]]
## [1] "Clifford"
## 
## [[5]]
## [1] "Professor" "Johns"    
## 
## [[6]]
## [1] NA
## 
## [[7]]
## [1] NA

Predicate Verb

get_phrase_type_regex(x, "VP") %>%
    take() %>%
    get_phrase_type_regex("(VB|MD)") %>%
    take() %>%
    get_leaves()

## [[1]]
## [1] "like"
## 
## [[2]]
## [1] "are"
## 
## [[3]]
## [1] "is"
## 
## [[4]]
## [1] "ate"
## 
## [[5]]
## [1] "can"
## 
## [[6]]
## [1] NA
## 
## [[7]]
## [1] NA

Direct Object

get_phrase_type_regex(x, "VP") %>%
    take() %>%
    get_phrase_type_regex("NP") %>%
    take() %>%
    get_leaves()

## [[1]]
## [1] "chocolate"
## 
## [[2]]
## NULL
## 
## [[3]]
## [1] "my"     "friend"
## 
## [[4]]
## [1] "my"    "lunch"
## 
## [[5]]
## NULL
## 
## [[6]]
## [1] NA
## 
## [[7]]
## [1] NA