An exploratory approach to the computational quantification of journalistic values

Published date11 February 2019
Pages133-148
DOIhttps://doi.org/10.1108/OIR-03-2018-0090
Date11 February 2019
AuthorSujin Choi
Subject MatterLibrary & information science,Information behaviour & retrieval,Collection building & management,Bibliometrics,Databases,Information & knowledge management,Information & communications technology,Internet,Records management & preservation,Document management
An exploratory approach to the
computational quantification of
journalistic values
Sujin Choi
Kyung Hee University, Seoul, Republic of Korea
Abstract
Purpose News algorithms not only help the authors to efficiently navigate the sea of available information,
but also frame information in ways that influence public discourse and citizenship. Indeed, the likelihood that
readers will be exposed to and read given news articles is structured into news algorithms. Thus, ensuring
that news algorithms uphold journalistic values is crucial. In this regard, the purpose of this paper is to
quantify journalistic values to make them readable by algorithms through taking an exploratory approach to
a question that has not been previously investigated.
Design/methodology/approach The author matched the textual indices (extracted from natural
language processing/automated content analysis) with human conceptions of journalistic values (derived
from survey analysis) by implementing partial least squares path modeling.
Findings The results suggest that the numbers of words or quotes news articles contain have a strong
association with the survey respondent assessments of their balance, diversity, importance and factuality.
Linguistic polarization was an inverse indicator of respondentsperception of balance, diversity and
importance. While linguistic intensity was useful for gauging respondentsperception of sensationalism, it
was an ineffective indicator of importance and factuality. The numbers of adverbs and adjectives were useful
for estimating respondentsperceptions of factuality and sensationalism. In addition, the greater numbers of
quotes, pair quotes and exclamation/question marks in news headlines were associated with respondents
perception of lower journalistic values. The author also found that the assessment of journalistic values
influences the perception of news credibility.
Research limitations/implications This study has implications for computational journalism,
credibility research and news algorithm development.
Originality/value It represents the first attempt to quantify human conceptions of journalistic values with
textual indices.
Keywords Digital journalism, Computational journalism, Credibility, Journalistic value, News algorithm
Paper type Research paper
The current information environment is subject to the attention economy (Lanham, 2006):
while our cognitive resources (i.e. attention) are limited, the information we can access is not.
In this circumstance, which information or news is produced is less important than
which information or news gains our attention is (Choi and Kim, 2017; Lotan, 2014).
News algorithms[1] help us allocate our attention to particular news or information through
the process of classification, association and prioritization (Diakopoulos, 2015). In brief,
algorithms classify news articles into certain categories (e.g. politics, business, technology,
science, sports, food and international affairs), form clusters on the basis of issue similarity
within particular categories and then prioritize the news articles included in each particular
cluster. This process ranks certain news articles higher than others and places them at the
top of the screen, where peoples attention usually remains (Nielsen, 2006; Sherman, 2005).
Therefore, the chance of news articles being exposed to and read by users is structured into
the news algorithms (Ananny, 2016; Gillespie, 2014).
Online Information Review
Vol. 43 No. 1, 2019
pp. 133-148
© Emerald PublishingLimited
1468-4527
DOI 10.1108/OIR-03-2018-0090
Received 15 March 2018
Revised 12 July 2018
4 October 2018
Accepted 5 October 2018
The current issue and full text archive of this journal is available on Emerald Insight at:
www.emeraldinsight.com/1468-4527.htm
This work was supported by a grant from Kyung Hee University in 2018 (KHU-20180926). In addition,
the author is grateful to the Korea Press Foundations News Trust Committee, of which the author was
a founding member, for helping inspire this study.
This paper forms part of a special section Social media mining for journalism.
133
Computational
quantification
News algorithmsarticle classification, association and prioritization processes resemble
the framing process traditionally used by journalists and legacy news organizations.
Framing defined as a process of persistent selection, emphasis, and exclusion
(Gitlin, 1980, p. 7) of perceived reality influences audience perception, organization and
interpretation of numerous pieces of information (Goffman, 1974). In controlling news
articlespotential exposure to online news users, news algorithms influence said users
perceptions of reality a process that has been labeled algorithmic reality construction
( Just and Latzer, 2017, p. 246). In addition to potentially affecting peoples perceptions of
reality, such reality construction may also affect democratic decision-making processes
(Bucher, 2017; Gillespie, 2014; Segev, 2017). In this regard, news algorithms not only help us
efficiently navigate the sea of available information, but also frame information in ways
that influence public discourse and citizenship.
This raises several important questions. How can we ensure that these publicly
important news algorithms (which select, rank and display news articles) uphold
journalistic values? Specifically, in what ways can we design these algorithms to understand
the notion of journalistic values and how can we quantify journalistic values? Except for the
developers and owners of news algorithms, no one currently has a clear answer regarding
the values or criteria news algorithms use; companies that own news algorithms are not yet
legally obliged to disclose this information[2]. However, increasing academic and public
demand for algorithmic transparency/accountability (e.g. Ananny, 2016; Diakopoulos, 2015;
Diakopoulos and Koliska, 2017; Dörr and Hollnbuchner, 2017; Just and Latzer, 2017;
Lotan, 2014; Oh, 2016; Oh and Kim, 2016; Stark and Diakopoulos, 2016) may lead to more
open discussion of the principles on which news algorithms are based, which could help
ensure that developers orient their algorithms to realize journalistic values. Concern
regarding covert news algorithms is already part of the public consciousness, as people
wonder: what happens when information providers no longer care about truth or
accuracy?(Sambrook, 2012, p. 10).
Furthermore, the negative side effects of the contemporary digital news environment
highlight the need for news algorithms that uphold journalistic values. The current digital
news environment has been described as a content farmor a digital sweatshop
(Bakker, 2012). To survive the so-called news cyclone,online news producers constantly
publish articles that have not gone through sufficient fact checking (Klinenberg, 2005).
The repetitive news phenomenon (Choi and Kim, 2017) and churnalism (Davies, 2009) also
prevail. News algorithms contribute to these phenomena by privileging more recent
and more search-word relevant news articles. To increase algorithmic choosabilityand
thereby bolster clickability,online news producers tend to churn out press releases,
become more repetitive, disregard fact checking processes and make stories more
provocative. This series of phenomena has fueled the decline of news credibility.
Acknowledging this context, this study took an exploratory approach to the
quantification of journalistic values (balance, variety, importance, factuality, readability
and sensationalism), based on the assumption that news articles that reflect journalistic
values are credible. Our quantification of journalistic values had two levels: the
quantification of human conceptions of journalistic values and the quantification of these
conceptions into natural language processing (NLP) indices that computers understand.
We reasoned that, for instance, the identification of news articles that people perceive as
balanced must precede the identification of the NLP indices that reflect human conceptions
of balance. Furthermore, people may not rely on a single journalistic value to assess the
credibility of specific news articles. Thus, we found it necessary to address the question of
which journalistic values explain news credibility.
In order to answer the questions above, we used the following methods: a survey of
human perception of journalistic values, the NLP on R platform, an automated content
134
OIR
43,1

To continue reading

Request your trial

VLEX uses login cookies to provide you with a better browsing experience. If you click on 'Accept' or continue browsing this site we consider that you accept our cookie policy. ACCEPT