Extracting Social Networks from Language Statistics

Sterling Hutchinson, Max Louwerse*

*Corresponding author for this work

Research output: Contribution to journalArticleScientificpeer-review

Abstract

Knowledge regarding social information is commonly believed to be derived from sources such as formal relationships and interviews and can be plotted as complex networks. We explored whether social networks can also be extracted through other means by using language statistics. In three computational studies we computed first-order and higher-order (latent semantic analysis) co-occurrences of story characters in three novels. These statistical linguistic frequencies entered in a multidimensional scaling analysis yielded a two-dimensional solution that correlated with the two-dimensional networks of characters generated by experts. An experimental study in which participants were asked to estimate social networks showed that human estimates are similar to computational estimates. These results demonstrated that language statistics based on texts can be used to generate social networks.

Original languageEnglish
Pages (from-to)607-618
Number of pages12
JournalDiscourse Processes
Volume55
Issue number7
DOIs
Publication statusPublished - 2018

Keywords

  • BIDIMENSIONAL REGRESSION
  • PERCEPTUAL SIMULATION
  • MERE EXPOSURE
  • ENCODES
  • MAPS
  • TIES

Cite this

Hutchinson, Sterling ; Louwerse, Max. / Extracting Social Networks from Language Statistics. In: Discourse Processes. 2018 ; Vol. 55, No. 7. pp. 607-618.
@article{238a2a48799c4e90814eac27d9c52106,
title = "Extracting Social Networks from Language Statistics",
abstract = "Knowledge regarding social information is commonly believed to be derived from sources such as formal relationships and interviews and can be plotted as complex networks. We explored whether social networks can also be extracted through other means by using language statistics. In three computational studies we computed first-order and higher-order (latent semantic analysis) co-occurrences of story characters in three novels. These statistical linguistic frequencies entered in a multidimensional scaling analysis yielded a two-dimensional solution that correlated with the two-dimensional networks of characters generated by experts. An experimental study in which participants were asked to estimate social networks showed that human estimates are similar to computational estimates. These results demonstrated that language statistics based on texts can be used to generate social networks.",
keywords = "BIDIMENSIONAL REGRESSION, PERCEPTUAL SIMULATION, MERE EXPOSURE, ENCODES, MAPS, TIES",
author = "Sterling Hutchinson and Max Louwerse",
year = "2018",
doi = "10.1080/0163853X.2017.1332446",
language = "English",
volume = "55",
pages = "607--618",
journal = "Discourse Processes",
issn = "0163-853X",
publisher = "ROUTLEDGE JOURNALS, TAYLOR & FRANCIS LTD",
number = "7",

}

Extracting Social Networks from Language Statistics. / Hutchinson, Sterling; Louwerse, Max.

In: Discourse Processes, Vol. 55, No. 7, 2018, p. 607-618.

Research output: Contribution to journalArticleScientificpeer-review

TY - JOUR

T1 - Extracting Social Networks from Language Statistics

AU - Hutchinson, Sterling

AU - Louwerse, Max

PY - 2018

Y1 - 2018

N2 - Knowledge regarding social information is commonly believed to be derived from sources such as formal relationships and interviews and can be plotted as complex networks. We explored whether social networks can also be extracted through other means by using language statistics. In three computational studies we computed first-order and higher-order (latent semantic analysis) co-occurrences of story characters in three novels. These statistical linguistic frequencies entered in a multidimensional scaling analysis yielded a two-dimensional solution that correlated with the two-dimensional networks of characters generated by experts. An experimental study in which participants were asked to estimate social networks showed that human estimates are similar to computational estimates. These results demonstrated that language statistics based on texts can be used to generate social networks.

AB - Knowledge regarding social information is commonly believed to be derived from sources such as formal relationships and interviews and can be plotted as complex networks. We explored whether social networks can also be extracted through other means by using language statistics. In three computational studies we computed first-order and higher-order (latent semantic analysis) co-occurrences of story characters in three novels. These statistical linguistic frequencies entered in a multidimensional scaling analysis yielded a two-dimensional solution that correlated with the two-dimensional networks of characters generated by experts. An experimental study in which participants were asked to estimate social networks showed that human estimates are similar to computational estimates. These results demonstrated that language statistics based on texts can be used to generate social networks.

KW - BIDIMENSIONAL REGRESSION

KW - PERCEPTUAL SIMULATION

KW - MERE EXPOSURE

KW - ENCODES

KW - MAPS

KW - TIES

U2 - 10.1080/0163853X.2017.1332446

DO - 10.1080/0163853X.2017.1332446

M3 - Article

VL - 55

SP - 607

EP - 618

JO - Discourse Processes

JF - Discourse Processes

SN - 0163-853X

IS - 7

ER -