fst_format: Annotate open-ended survey responses in into CoNLL-U format

In finnsurveytext: Analyse Open-Ended Survey Responses in Finnish

Annotate open-ended survey responses in into CoNLL-U format

Description

Creates a dataframe in CoNLL-U format from a dataframe containing text from using the [udpipe] package and a language model plus any additional columns that are included such as 'weights' or columns added through 'add_cols'.

Usage

fst_format(data, question, id, model = "ftb", weights = NULL, add_cols = NULL)

Arguments

data	A dataframe of survey responses which contains an open-ended question.
question	The column in the dataframe which contains the open-ended question.
id	The column in the dataframe which contains the ids for the responses.
model	A language model available for [udpipe]. '"ftb"' (default) or '"tdt"' are recognised as shorthand for "finnish-ftb" and "finnish-tdt". The full list is available in the [udpipe] documentation or via 'fst_print_available_models()'.
weights	Optional, the column of the dataframe which contains the respective weights for each response.
add_cols	Optional, a column (or columns) from the dataframe which contain other information you'd like to retain (for instance, covariate columnns for splitting the data for comparison plots).

Value

Dataframe of annotated text in CoNLL-U format plus any additional columns.

Examples

## Not run: 
i <- "fsd_id"
fst_format(data = child, question = "q7", id = i)
fst_format(data = child, question = "q7", id = i, model = "tdt")
fst_format(data = child, question = "q7", id = i, weights="paino")
cols <- c("gender", "major_region", "daycare_before_school")
fst_format(child, question = "q7", id = i, add_cols = cols)
fst_format(child, question = "q7", id = i, add_cols = "gender, major_region")
fst_format(child, question = 'q7', id = i, model = 'swedish-talbanken')
unlink("finnish-ftb-ud-2.5-191206.udpipe")
unlink("finnish-tdt-ud-2.5-191206.udpipe")
unlink("swedish-talkbanken-ud-2.5-191206.udpipe")

## End(Not run)

finnsurveytext documentation built on April 4, 2025, 5:07 a.m.