s_attributes: Using Structural Attributes.

In RcppCWB: 'Rcpp' Bindings for the 'Corpus Workbench' ('CWB')

Using Structural Attributes.

Description

Structural attributes store the metadata of texts in a CWB corpus and/or any kind of annotation of a region of text. The fundamental unit are so-called strucs, i.e. indices of regions identified by a left and a right corpus position. The corpus library (CL) offers a set of functions to make the translations between corpus positions (cpos) and strucs (struc).

Usage

cl_cpos2struc(
  corpus,
  s_attribute,
  cpos,
  registry = Sys.getenv("CORPUS_REGISTRY")
)

cl_struc2cpos(
  corpus,
  s_attribute,
  registry = Sys.getenv("CORPUS_REGISTRY"),
  struc
)

cl_struc2str(
  corpus,
  s_attribute,
  struc,
  registry = Sys.getenv("CORPUS_REGISTRY")
)

cl_cpos2lbound(corpus, s_attribute, cpos, registry = NULL)

cl_cpos2rbound(corpus, s_attribute, cpos, registry = NULL)

Arguments

corpus	name of a CWB corpus (upper case)
s_attribute	name of structural attribute (character vector)
cpos	An integer vector with corpus positions.
registry	path to the registry directory, defaults to the value of the environment variable CORPUS_REGISTRY
struc	a struc identifying a region

Details

cl_cpos2rbound() and cl_cpos2lbound() return NA for values of cpos that are outside a struc for the structural attribute given.

Examples

# get metadata for matches of token
# scenario: id of the texts with occurrence of 'oil'
token_to_get <- "oil"
token_id <- cl_str2id("REUTERS", p_attribute = "word", str = "oil", get_tmp_registry())
token_cpos <- cl_id2cpos("REUTERS", p_attribute = "word", id = token_id, get_tmp_registry())
strucs <- cl_cpos2struc("REUTERS", s_attribute = "id", cpos = token_cpos, get_tmp_registry())
strucs_unique <- unique(strucs)
text_ids <- cl_struc2str("REUTERS", s_attribute = "id", struc = strucs_unique, get_tmp_registry())

# get the full text of the first text with match for 'oil'
left_cpos <- cl_cpos2lbound(
  "REUTERS", s_attribute = "id",
  cpos = min(token_cpos),
  registry = get_tmp_registry()
)
right_cpos <- cl_cpos2rbound(
  "REUTERS",
  s_attribute = "id",
  cpos = min(token_cpos),
  registry = get_tmp_registry()
)
txt <- cl_cpos2str(
  "REUTERS", p_attribute = "word",
  cpos = left_cpos:right_cpos,
  registry = get_tmp_registry()
)
fulltext <- paste(txt, collapse = " ")

# alternativ approach to achieve same result
first_struc_match_oil <- cl_cpos2struc(
  "REUTERS", s_attribute = "id",
  cpos = min(token_cpos),
  registry = get_tmp_registry()
  )
cpos_struc <- cl_struc2cpos(
  "REUTERS", s_attribute = "id",
  struc = first_struc_match_oil,
  registry = get_tmp_registry()
)
txt <- cl_cpos2str(
  "REUTERS",
  p_attribute = "word",
  cpos = cpos_struc[1]:cpos_struc[2],
  registry = get_tmp_registry()
)
fulltext <- paste(txt, collapse = " ")

RcppCWB documentation built on July 9, 2023, 7:40 p.m.