Coherence#

The coherence components calculates the coherence of the document, based on word embedding cosine similarity between sentences.

textdescriptives currently implements first-order and second-order coherence. The implementation follows e.g. [1] and [2]: * First-order coherence: The cosine similarity between consecutive sentences. * Second-order coherence: The cosine similarity between sentences that are two sentences apart.

The implementation uses spacy’s Span.similarity method to calculate the similarity between sentences. The similarity is based on the word embeddings in the spacy pipeline, i.e. small, medium, large, or transformer models will have different results. If you want to use a specific word embedding (e.g. fasttext) you should overwrite the Doc.vector attribute. Read more on spacy’s documentation for similarity and overwriting the vector attribute

The following attributes are added to Doc objects.:

  • ._.first_order_coherence_values: A list of floats, where each float is the cosine similarity between consecutive sentences.

  • ._.second_order_coherence_values: A list of floats, where each float is the

    cosine similarity between sentences that are two sentences apart.

  • ._.cohererence: a dict containing the mean coherence values for first and second order coherence (keys: “first_order_coherence”, “second_order_coherence”)

[1] Bedi, G., Carrillo, F., Cecchi, G. A., Slezak, D. F., Sigman, M., Mota, N. B., Ribeiro, S., Javitt, D. C., Copelli, M., & Corcoran, C. M. (2015). Automated analysis of free speech predicts psychosis onset in high-risk youths. Npj Schizophrenia, 1(1), Article 1. https://doi.org/10.1038/npjschz.2015.30

[2] Parola, A., Lin, J. M., Simonsen, A., Bliksted, V., Zhou, Y., Wang, H., Inoue, L., Koelkebeck, K., & Fusaroli, R. (2022). Speech disturbances in schizophrenia: Assessing cross-linguistic generalizability of NLP automated measures of coherence. Schizophrenia Research. https://doi.org/10.1016/j.schres.2022.07.002

Usage#

import spacy
import textdescriptives as td
nlp = spacy.load("en_core_web_lg")
nlp.add_pipe("textdescriptives/coherence")
doc = nlp("The world is changed. I feel it in the water. I feel it in the earth. I smell it in the air. Much that once was is lost, for none now live who remember it.")

# all attributes are stored as a dict in the ._.coherence attribute
doc._.coherence

# first and second order coherence values are also added as separate attributes
# note that the first/first two sentences do not have any values for first/second order
# coherence, respectively, as they require 1 or 2 sentences to average over
doc._.first_order_coherence_values
doc._.second_order_coherence_values

# extract to dataframe
td.extract_df(doc)

text

first_order_coherence

second_order_coherence

0

The world is changed(…)

0.780735

0.749475

Component#

textdescriptives.components.coherence.create_coherence_component(nlp: Language, name: str) → Callable[[Doc], Doc][source]#

Allows Coherence to be added to a spaCy pipe using nlp.add_pipe(“textdescriptives/coherence”).

Adding this component to a pipeline sets the following attributes:

  • doc._.first_order_coherence_values

  • doc._.second_order_coherence_values

  • doc._.coherence

Parameters:

  • nlp (Language) – spaCy language object, does not need to be specified in the nlp.add_pipe call.

  • name (str) – name of the component. Can be optionally specified in the nlp.add_pipe call, using the name argument.

Returns:

The Coherence component to be added to the pipe.

Return type:

Callable[[Doc], Doc]

Examples

>>> import spacy
>>> import textdescriptives as td
>>> nlp = spacy.load("en_core_web_md")
>>> nlp.add_pipe("textdescriptives/coherence")
>>> # apply the pipeline to a text
>>> doc = nlp("This is a sentence. This is another sentence.")
>>> # get coherence values
>>> doc._.coherence