| Title: | Directness and Intensity of Conflict Expression |
|---|---|
| Description: | A Natural Language Processing Model trained to detect directness and intensity during conflict. See <https://www.mikeyeomans.info>. |
| Authors: | Michael Yeomans [aut, cre] |
| Maintainer: | Michael Yeomans <[email protected]> |
| License: | MIT + file LICENSE |
| Version: | 0.1.0 |
| Built: | 2025-02-18 05:43:10 UTC |
| Source: | https://github.com/myeomans/dicem |
Simple features as inputs to the DICE model
basicSet(text)basicSet(text)
text |
character A vector of texts, each of which will be tallied for DICE features. |
The DICE models use, as features, linear and quadratic terms for sentiment, emotion, and word count.
a data.frame of feature scores for the pre-trained models.
Detects linguistic markers of politeness in natural language. Takes an N-length vector of text documents and returns an N-row data.frame of scores on the two DICE dimensions.
DICE(text, parser = c("none", "spacy"), uk_english = FALSE, num_mc_cores = 1)DICE(text, parser = c("none", "spacy"), uk_english = FALSE, num_mc_cores = 1)
text |
character A vector of texts, each of which will be tallied for DICE features. |
parser |
character Name of dependency parser to use (see details). Without a dependency parser, some features will be approximated, while others cannot be calculated at all. |
uk_english |
logical Does the text contain any British English spelling? Including variants (e.g. Canadian). Default is FALSE |
num_mc_cores |
integer Number of cores for parallelization. Default is 1, but we encourage users to try parallel::detectCores() if possible. |
The best intensity model uses politeness features, which depend on part-of-speech tagged sentences (e.g. "bare commands" are a particular verb class). To include these features in the analysis, a POS tagger must be initialized beforehand - we currently support SpaCy which must be installed separately in Python (see example for implementation). If not, a simpler model can be used, though it is somewhat less accruate.
a data.frame of scores on directness and intensity.
Weingart et al., 2015 Yeomans et al., 2024
data("phone_offers") DICE(phone_offers$message[1:10], parser="none") ## Not run: # Detect multiple cores automatically for parallel processing DICE(phone_offers$message, num_mc_cores=parallel::detectCores()) # Connect to SpaCy installation for part-of-speech features # THIS REQUIRES SPACY INSTALLATION OUTSIDE OF R # For some machines, spacyr::spacy_install() will work, but please confirm before running spacyr::spacy_initialize(python_executable = PYTHON_PATH) DICE(phone_offers$message, parser="spacy") ## End(Not run)data("phone_offers") DICE(phone_offers$message[1:10], parser="none") ## Not run: # Detect multiple cores automatically for parallel processing DICE(phone_offers$message, num_mc_cores=parallel::detectCores()) # Connect to SpaCy installation for part-of-speech features # THIS REQUIRES SPACY INSTALLATION OUTSIDE OF R # For some machines, spacyr::spacy_install() will work, but please confirm before running spacyr::spacy_initialize(python_executable = PYTHON_PATH) DICE(phone_offers$message, parser="spacy") ## End(Not run)
For internal use only. This dataset demonstrates the ngram features that are used for the pre-trained models.
diceNGramsdiceNGrams
A (truncated) matrix of ngram feature counts for alignment to the pre-trained glmnet models.
Yeomans et al., (2024). A Natural Language Processing Model for Conflict Expression in Conversation
Extracts feature sets to match pre-trained models
featureSet(text, parser = c("none", "spacy"), num_mc_cores = 1)featureSet(text, parser = c("none", "spacy"), num_mc_cores = 1)
text |
character A vector of texts, each of which will be tallied for politeness features. |
parser |
character Name of dependency parser to use (see details). Without a dependency parser, the politeness features are excluded from the model. |
num_mc_cores |
integer Number of cores for parallelization. Default is 1, but we encourage users to try parallel::detectCores() if possible. |
The politeness features depend on part-of-speech tagged sentences (e.g. "bare commands" are a particular verb class). To include these features in the analysis, a POS tagger must be initialized beforehand - we currently support SpaCy which must be installed separately in Python (see example for implementation).
a data.frame of features, matching the pre-trained model set
A dataset containing the purchase offer message and a label indicating if the writer was assigned to be warm (1) or tough (0)
phone_offersphone_offers
A data frame with 355 rows and 2 variables:
character of purchase offer message
binary label indicating if message is warm or tough
Jeong, M., Minson, J., Yeomans, M. & Gino, F. (2019).
"Communicating Warmth in Distributed Negotiations is Surprisingly Ineffective."
Study 1. https://osf.io/t7sd6/
This calculates the polynomial projection of the simple features used during model training
polymodelpolymodel
A pre-trained polynomial equation
For internal use only. This dataset contains a quanteda dictionary for converting UK words to US words. The models in this package were all trained on US English.
uk2usuk2us
A quanteda dictionary with named entries. Names are the US version, and entries are the UK version.
Borrowed from the quanteda.dictionaries package on github (from user kbenoit)