edstr_view(): interactive exploration • edstr

library(edstr)

edstr_view() searches for regex patterns in a text column and displays matching tokens, counts, and highlighted text. Unlike other pipeline functions, it does not save any files; it is designed for fast, iterative pattern development before running edstr_extract().

Prerequisites

edstr_config(
  edstr_dirname = "output/my_study",
  edstr_filename = "my_study",
  edstr_text = "note_text",
  edstr_overwrite = FALSE
)

df_import <- edstr_import()
df_clean <- edstr_clean(data = df_import, replace = c("\\s+" = " "))

Basic usage

Pass a regex pattern to search for in the text column.

result <- edstr_view(
  data = df_clean,
  pattern = "fractur"
)

The function prints a frequency table of distinct matches and a summary of match counts, then invisibly returns a list with three elements:

match: a tibble of all matches (one row per match, with the document ID)
count: a tibble of distinct matches sorted by frequency
text: highlighted output from stringr::str_view()

Capturing multi-word expressions

The ngrams argument sets the width of the regex window captured, the matched word included: edstr_view() does not tokenise, it widens pattern by up to ngrams - 1 further words. This is useful for discovering common multi-word patterns. It is a display window, not a search setting: the n-gram sizes edstr_extract() searches are set by its own ngram_max argument.

result <- edstr_view(
  data = df_clean,
  pattern = "fractur",
  ngrams = 3
)

With ngrams = 3, a match on "fractur" captures up to two additional words, e.g. "fracture col femoral" or "fracture extremite superieure".

Applying cleaning rules inline

The replace argument accepts the same format as edstr_clean() and applies replacements before matching. This lets you test cleaning rules without modifying the data.

result <- edstr_view(
  data = df_clean,
  pattern = "diabete",
  replace = c("diabète" = "diabete"),
  ngrams = 2
)

Typical workflow

A common pattern is to iterate with edstr_view() until the regex captures the right matches, then use the finalised pattern in edstr_extract():

Start broad: edstr_view(data, pattern = "diabet", ngrams = 3)
Review the count table to identify common expressions and false positives
Refine the pattern or add exclusions
Once satisfied, pass the pattern as a concept to edstr_extract()