CreateTextFeatures: Creates text-based features from a radiology report corpus

In wlktan/LireNLPSystem: LireNLPSystem

Description

This function creates N-gram features from a radiology report corpus, where N-gram

Usage

CreateTextFeatures(segmented.reports, id_col = "imageid",
  text.cols = c("body", "impression"),
  all.stop.words = setdiff(stopwords(), c("no", "not", "nor")),
  finding.dictionary = NULL, docfreq = "prop", min_doc_prop = 0,
  max_doc_prop = 1, termfreq = "count", min_term_freq = 1,
  max_term_freq = NULL, tf_type = "boolean", df_type = "unary",
  n_gram_length = 1)

Arguments

segmented.reports	Input data frame with
id_col	The ID column in segmented.reports, defaults to imageid
text.cols	Vector of findings text column names in segmented.reports, defaults to c("body","impression")
all.stop.words	List of stop words, defaults to English stopword list excluding negation
finding.dictionary	Dictionary object to map findings, defaults to NULL
docfreq	See quanteda::dfm_trim; One of "count", "inverse", "inversemax", "inverseprob", "unary"
min_doc_prop	minimum/maximum values of a feature's document frequency, below/above which features will be removed
termfreq	See quanteda::dfm_trim; One of "count", "prop", "propmax", "logcount", "boolean", "augmented", "logave"
min_term_freq	minimum/maximum values of feature frequencies across all documents, below/above which features will be removed
max_term_freq	Above
tf_type	See quanteda::dfm_weight; One of "count", "prop", "propmax", "logcount", "boolean", "augmented", "logave"
df_type	See quanteda::docfreq; One of "count", "inverse", "inversemax", "inverseprob", "unary"
n_gram_length	Unigram, bigram, or trigram features; defaults to 3 (trigrams)

Value

A document frequency matrix with each row as a unique report, each column is a feature, and the cells are the counts in the document.

Examples

CreateTextFeatures(segmented.reports)

wlktan/LireNLPSystem documentation built on May 27, 2019, 12:13 p.m.