Elsevier

Expert Systems with Applications

Volume 40, Issue 11, 1 September 2013, Pages 4600-4608
Expert Systems with Applications

Complex network tools in building expert systems that perform framing analysis

https://doi.org/10.1016/j.eswa.2013.01.064Get rights and content

Abstract

Framing, in its specific application to media research, is defined as the “central organizing idea for making sense of an issue or conflict and suggesting what is at stake.” It can be found in various disciplines of the social sciences, most notably in political science, psychology, and communication research. Due to the fuzzy nature of frames, identifying them has proven to be quite complex. Here, we perform framing analysis on a corpus of news texts on the population and family planning issue in the Philippines by operating two varying approaches: human-based and computer-assisted. A singular holistic approach to framing is initially implemented where coders/domain experts classify each news text to a specific pre-defined frame. This traditional approach is known to raise serious issues on the reliability and validity of the results mainly due to human’s intrinsic biases. To address such issues, we propose a novel technique that synergically combines the method of Matthes and Kohring (2008) and complex networks approach. In our model, the codings of texts are cast as a network of content analytic variables (CAVs). Our proposed method tackles the clustering issue that MK raised, which plagues framing scholars in the quantitative identification of news frames in texts. Moreover, the research is significant on a societal level as it also aims to gain perspective for reasons on the lack of progress in discussions about suitable population policies in most developing countries like the Philippines.

Highlights

► We demonstrate a novel framing analysis technique based on complex networks approach. ► The proposed procedure is compared with human-based and syntactic network analysis. ► Consistent frames are extracted using a corpus of news texts on population & family planning issues. ► Our method address clustering problems in the quantitative identification of news frames in texts. ► Our results can explain the lack of progress in discussions about suitable population policies.

Introduction

In recent years, much attention has been given to the study of frames in the field of communication research, particularly focusing on how the news media frame issues in text. Framing analysis is salient in understanding the potential of the media to influence public opinion (Entman, 1991, Bardhan, 2001). The subject has garnered a huge appeal across the communication and media disciplines including, but not limited to, policy, media content, and cultural studies (Joachim, 2003, Shah et al., 2002, Hoerl et al., 2009, Yeo et al., 2007, Ryan et al., 2001).

Framing, in this context, is interpreted as the ”central organizing idea for making sense of an issue or conflict and suggesting what is at stake (Gamson, 1989, Koch, 1998).” The main challenge is to identify and asses the latent schemes that emerge from a reporting of an issue and recognize them in the most optimized, reliable, and accurate way.

Early framing studies were particularly qualitative in nature using entirely hermeneutic approaches. As such, most of these analyzes were highly dependent on the experts doing the exegesis. Hence, such subjective procedure could potentially cast doubts on the reliability and accuracy of the interpretation (Scheufele and Scheufele, 2010, Van Gorp, 2005); moreover, issues on replicability plague the evaluation measures.

More recent studies, on the other hand, address these issues using more exact quantitative models (Miller, 1997, Miller et al., 1998). Quantitative framing researchers argue that the methods provided are more precise and straightforward and are repeatable – usually done with the aide of “devices” as frame indicators (Koella, 2003). Most of these techniques are computer-assisted, which is the exact antithesis of the more traditional approach. Typically, the methods utilize frequency-statistics of certain keywords and their loci in the body of the texts (Legara et al., 2010, Murphy, 2001). Although the approach improves reliability, criticisms regarding its validity have been raised since it has been shown that some infrequently occurring words in the text could actually be “central to the meaning of a text”; and, by filtering in only the most frequently occurring words in the examination, a significant amount of nontrivial information can be lost in the process (Scheufele and Scheufele, 2010, Matthes and Kohring, 2008).

Although much has been done in expanding its domain, existing risks vis-à-vis the reliability and validity of framing protocols are still of major concern. Due to this fuzzy nature of frames, naming and quantifying them have proven to be quite complex, which calls for a multi-disciplinary approach involving both domain and computation experts.

On the other hand, the science of networks has emerged to be a leading approach in probing signatures of complex systems inherent in both nature and society. Networks have been shown to detect cryptic patterns that contain germane information about such systems, e.g. DNA nuecleotides sequence data (Sinatra, Condorelli, & Latora, 2010), dissimilarity between poem and prose (Roxas & Tapang, 2010). In this study, one of the main challenges in applying network theory is in deciphering the hidden patterns through the use of the most appropriate symbolic representations of the compendium of data. It is important to correctly detect the “fundamental units carrying information” in constructing the networks (Sinatra et al., 2010).

Here, we investigate and improve upon existing framing analysis protocols and tackle them in the context of network theory. The research is relevant on two levels: social and methodological. The procedures performed here utilize text from media coverage of a vehemently debated issue in most, if not all, developing countries – the issue on population, reproductive health, and family planning. The population issue has consternated governments of developing countries when concerns are retracted to issues on poverty and women’s health. Particularly, we look into the Philippine news coverage. The Philippines is a developing country that is predominantly Catholic (80% of the population), and is ranked fourth (4) worldwide in terms of total population and population density and has been noted by the United States Census Bureau as a fast growing country (Central Intelligence Agency Online Factbook, 2012).

On the methodological significance, detecting media frames has produced a significant number of theories, techniques, and procedures. However, there are very few studies that compares the consistency of the results generated by these various methods because typically only one framing procedure is used to dissect an issue. Here, we utilize and compare three distinct methods in the study of media framing and compare their consistencies, advantages and procedural differences.

Section snippets

Data

This study covers a total of 346 news articles related to issues on population, family planning, reproductive health, and contraception. The period covered runs from 1987 to 2007. Data were taken from three of the most widely circulated broadsheets in the Philippines namely, the Philippine Daily Inquirer (PDI), the Philippine Star (PS), and the Manila Bulletin (MB).

Singular holistic approach to frames

Perhaps the most traditional method used to find frames is through the perusal of texts by experts of both the issues under study and the field of communication research. A singular holistic approach (SHA) to framing, which is deductive in nature, was carried out. Here, a single frame was considered as a lone reference to a whole article. Coders defined six frames after methodically familiarizing the topic through a careful review of a small sample of text on the issue. The pre-defined frame

Frames as frame elements approach

In Section 4, frames are treated as a collection of elements comprising an issue, which makes the analysis more guided. Essentially, frames are dissected into several parts that are interconnected to each other. This notion was first introduced by Kohring and Matthes (2002) and the conceptual basis behind the technique stemmed from the widely-accepted definition of framing given by Entman: “the process of highlighting some select aspects of a news or an article to publicize a specific problem

Results and discussion

The results using the community detection algorithm (CDA) indicate that the most optimal partition would be to group the articles into three (3) communities (see Table 8), with Q = 0.1550 for several runs, which indicates that the number of edges within groups exceeds the number expected on the basis of chance. These patterns essentially descirbe the co-occurrences and interconnectivity of the binary variables in the corpus. Moreover, since frames consist of such patterns, the three communities

Acknowledgments

The authors acknowledge help of J.M. Atun in the data collection and coding process. C.M. and E.F.L are supported by Singapore ASTAR SERC Complex Systems Programme’s research grant (1224504056).

References (35)

  • M.R. Forster

    Key concepts in model selection: Performance and generalizability

    Journal of Mathematical Psychology

    (2000)
  • H. Akaike

    Information theory and an extension of the maximum likelihood principle

  • N. Bardhan

    Translational AIDS–HIV news narratives: A critical exploration of overarching frames

    Mass Communication and Society

    (2001)
  • V.D. Blondel et al.

    Fast unfolding of communities in large networks

    Journal Statistical Mechanics: Theory and Experiment

    (2008)
  • Central Intelligence Agency Online Factbook. ISSN 1553-8133. Website:...
  • C. David et al.

    Finding frames: Comparing two methods of frame analysis

    Communication Methods and Measures

    (2011)
  • R.M. Entman

    Framing U.S. coverage of international news: Contrasts in narratives of the KAL and Iran air incidents

    Journal of Communication

    (1991)
  • R.M. Entman

    Framing: Toward clarification of a fractured paradigm

    Journal of Communication

    (1993)
  • W.A. Gamson

    News as framing: Comment on Graber

    American Behavioral Scientist

    (1989)
  • K.E. Hoerl et al.

    Deranged loners and demented outsiders? Therapeutic news frames of presidential assassination attempts, 1973-2001

    Communication, Culture & Critique

    (2009)
  • J. Joachim

    Framing issues and seizing opportunities: The UN, NGOs, and womens rights

    International Studies Quarterly

    (2003)
  • E.E. Jones et al.

    Foundations of Social Psychology

    (1967)
  • J.W. Koch

    Political rhetoric and political persuasion: The changing structure of citizens preferences on health insurance during policy debates

    Public Opinion Quarterly

    (1998)
  • Koella, J. C., (2003). The impact of interest groups and news media framing on public opinion: The rise and fall of the...
  • M. Kohring et al.

    The face(t)s of biotech in the nineties: How the German press framed modern biotechnology

    Public Understanding of Science

    (2002)
  • P.F. Lazarsfeld

    The logical and mathematical foundations of latent structure analysis

  • E. Legara et al.

    News framing of population and family planning issues via syntactic network analysis

    International Journal of Modern Physics C

    (2010)
  • Cited by (7)

    • Executive summaries of uncertain values close to the gain/loss threshold – linguistic modelling perspective

      2020, Expert Systems with Applications
      Citation Excerpt :

      It is also known that the framing of the problem, or its parts, can significantly affect the path taken to its solution, the risk attitude of the decision-maker (Thaler, Tversky, Kahneman, & Schwartz, 1997) or even the overall interpretation of the situation. As Legara, Monterola, and David (2013) point out the framing can prove to be of crucial importance also in the design of expert systems and analytical models. Framing (in its general understanding of “making sense of the problem”) can even be applied within expert systems, where if done correctly, it can help to carry out the analysis more efficiently or using the most appropriate methods.

    • Disease spreading in complex networks: A numerical study with Principal Component Analysis

      2018, Expert Systems with Applications
      Citation Excerpt :

      However, depending on the problem, population may need a proper mathematical tool to consider space as an important factor, like cellular automata (Holko, Mdrek, Pastuszak, & Phusavat, 2016). More especifically, complex network approaches have proven to be a suitable tool for building expert systems, most notably in social sciences (Legara, Monterola, & David, 2013; Wachs-Lopes & Rodrigues, 2016). In general, complex network architecture is used to build and evaluate prediction models.

    • Improving data partition schemes in Smart Grids via clustering data streams

      2014, Expert Systems with Applications
      Citation Excerpt :

      This approach is based on algorithms that automatically explore data in order to uncover subjacent patterns. A cornerstone of unsupervised learning is found in data clustering, which consists in grouping examples into sets—the clusters—according to some proximity criterion with the aim of searching for hidden patterns (García-Piquer, 2012; Khan & Ahmad, 2013; Legara, Monterola, & David, 2013). The concept of making clusters out of input data is extrapolated to data streams in the field of clustering data streams.

    • News frames of the population issue in the Philippines

      2014, International Journal of Communication
    View all citing articles on Scopus
    View full text