politenessStrategies

Computes Politeness Strategies features.

Currently offering three strategy collections covering two languages:

Example usage:

class convokit.politenessStrategies.politenessStrategies.PolitenessStrategies(parse_attribute_name: str = 'parsed', strategy_attribute_name: str = 'politeness_strategies', marker_attribute_name: str = 'politeness_markers', strategy_collection: str = 'politeness_api', verbose: int = 0)

Encapsulates extraction of politeness strategies from utterances in a Corpus.

Parameters

  • parse_attribute_name – metadata attribute name to read parses from. Default is ‘parsed’.

  • strategy_attribute_name – metadata attribute name to store politeness strategies features under during the transform() step. Default is ‘politeness_strategies’.

  • marker_attribute_name – metadata attribute name to store politeness markers under during the transform() step. Default is ‘politeness_markers’.

  • strategy_collection – collection of politeness strategies to extract. Options include: “politeness_api”: English politeness strategies proposed in A computational approach to politeness with application to social factors (https://www.cs.cornell.edu/~cristian/Politeness.html) “politeness_local”: English politeness strategies realized through local markers as used in Facilitating the Communication of Politeness through Fine-Grained Paraphrasing (https://www.cs.cornell.edu/~cristian/Politeness_Paraphrasing.html) “politeness_cscw_zh”: Chinese politeness strategies adapted from `Studying Politeness across Cultures using English Twitter and Mandarin Weibo (https://dl.acm.org/doi/abs/10.1145/3415190) Default is “politeness_api”.

  • verbose – whether and how often to print status messages while computing features.

transform(corpus: convokit.model.corpus.Corpus, selector: Optional[Callable[[convokit.model.utterance.Utterance], bool]] = <function PolitenessStrategies.<lambda>>, markers: bool = False)

Extract politeness strategies from each utterances in the corpus and annotate the utterances with the extracted strategies. Requires that the corpus has previously been transformed by a Parser, such that each utterance has dependency parse info in its metadata table.

Parameters

  • corpus – the corpus to compute features for.

  • selector – a (lambda) function that takes an Utterance and returns a bool indicating whether the utterance should be included in this annotation step.

  • markers – whether or not to add politeness occurrence markers

transform_utterance(utt: convokit.model.utterance.Utterance, spacy_nlp: Callable[[str], spacy.tokens.Doc] = None, markers: bool = False)

Extract politeness strategies for raw string inputs (or individual utterances)

Parameters

utt – the utterance to be annotated with politeness strategies.

Spacy_nlp

if provided, will use this SpaCy object to do parsing; otherwise will initialize an object via load(‘en’).

Returns

the utterance with politeness annotations.

summarize(corpus: convokit.model.corpus.Corpus, selector: Callable[[convokit.model.utterance.Utterance], bool] = <function PolitenessStrategies.<lambda>>, plot: bool = False, y_lim=None)

Calculates strategy prevalence and plot graph if plot == True, with an optional selector that specifies which utterances to include in the analysis.

Parameters

  • corpus – the target Corpus

  • selector – a function (typically, a lambda function) that takes an Utterance and returns True or False (i.e. include / exclude).

By default, the selector includes all Utterances in the Corpus. :param plot: whether or not to output graph. :return: a pandas DataFrame of scores with graph optionally outputted