Overview of Computational Communication Research

Computational Communication Research (CCR): Overview

Introduction

Computational communication research (CCR) has emerged prominently as a sector within computational social science (CSS), offering innovative methods to tackle both traditional and new inquiries in communication studies. This paper examines the development and methodological insights of CCR within the global research landscape and particularly in mainland China. Key computational methods discussed include:

  • User Analytics

  • Content Mining

  • Computational Experiments

Keywords

  • Computational Social Science

  • Computational Communication

  • User Analytics

  • Content Mining

  • Computational Experiments

The Rise and Growth of CCR Around the World

Establishment of CCR

  • CCR is a subset of CSS, initiated by the CSS manifesto's publication in 2009 (Lazer et al., 2009).

  • Previous efforts in using computational methods in the social sciences were largely isolated and breadth-focused, encompassing terms like "social computing" and "internet social science."

Collaborative Foundations

  • Established a long-term partnership between the Web Mining Lab at City University of Hong Kong and the Tianwang Lab at Peking University starting in 2005 to explore Chinese websites (Zhu et al., 2008).

  • Advocacy for interdisciplinary collaboration between social sciences and computer science resulted in works like:

    • “Let the Social Sciences Ride on the Information Technology (IT) Bullet Train” (Li and Zhu, 2006)

    • “An Easy and Affordable Tool for e-Social Science Research” (Zhu and Li, 2007a)

  • The collaborative engagement evolved into the term “e-social science” (eSS), enhancing its global stature through Wikipedia and prominent publications.

Growth in CCR

  • Following the CSS manifesto, many researchers adopted CSS, reinforcing it as a unifying framework.

  • CCR's inception was documented in Zhu et al. (2014), which used the 5W communication questions (who, what, to whom, through which channels, with what effects) to structure research inquiries.

Journals and Conferences

  • The establishment of the Computational Methods Interest Group within the International Communication Association (ICA) in 2016 fostered rapid growth, with its evolution into a division once membership surpassed 200.

  • Launched the journal Computational Communication Research in 2019.

  • The Chinese Association of Computational Communication Research (CACCR) was founded in 2018 as a similar initiative.

Factors Contributing to the Adoption of CSS in Communication Research

1. Adaptation to New Media

  • Historically, communication researchers promptly responded to technological innovations such as radio, television, and online media, recognizing the potential of big data analytics early through user logs and ratings data.

2. Interdisciplinary Nature of Communication Research

  • The field's overlap with information and networking technology necessitated computing skills and access to resources, fostering early adoption of computational methods.

3. Intellectual Connections

  • Overlaps between traditional quantitative approaches (like probability sampling and causal inference) and new computational technologies facilitated smoother transitions into CCR methodologies.

Overview of CCR Methodology

Traditional vs. Computational Methods
Traditional Quantitative Methods

  • Prior forms predominantly used survey, experiment, and content analysis focusing on the 5W inquiries.

Computational Methods

  • Post-CSS emergence, communication researchers have utilized:

    • User Analytics

    • Content Mining

    • Computational Experiments

Comparative Overview (Table 1: Mapping Quantitative and Computational Methods)

  • Content Analysis (Quantitative) vs. Content Mining (Computational)

    • Content Analysis: Labor-intensive, human-operated data processing with high trust.

    • Content Mining: Automated data processing, higher efficiency but potential trust issues.

  • Similar structures apply for Survey vs. User Analytics and Experiments vs. Computational Experiments.

User Analytics

Definition

  • User analytics refers to methods of collecting and analyzing user behavior on social media using log data (Zhu et al., 2019).

User Logs Data

  • Comprises detailed recordings of social media interaction, such as timestamps, posting frequency, content types, and engagement metrics.

  • Origin of online user analytics traces back to pre-internet audience metrics via self-reported surveys in earlier decades, transitioning to automated viewer metrics like the "Peoplemeter" in television.

Comparison with Traditional Surveys and Content Mining
Survey vs. User Analytics

  • Surveys: Relies on self-reported behavior, subject to biases and errors.

  • User Analytics: Offers direct logs of user actions for more accurate behavioral interpretations.

User Analytics vs. Content Mining

  • User Analytics: Focused on behavioral data from interactions.

  • Content Mining: Analyzes user-generated content (UGC) and has its limitations due to the public nature of UGC.

Data Collection Methods

  • Methods include purchasing data, downloading from archives, API retrieval, and web scraping. Each method varies in accessibility, permissions, and data utility (Liang and Zhu, 2017; Zhu et al., 2019).

Ethical and Legal Considerations

  • The imperative for compliance with legal and ethical standards in data collection.

User Analytics Methods Applied

  • Longitudinal methods enable trend analyses; commonly analyzed via multiple regression and structural equation modeling (SEM).

Empirical Applications of User Analytics

Application 1: Invisible Activities on Social Media

  • Benevenuto et al. (2009): Study on Orkut reveals over 90% were invisible activities like browsing, underscoring biases in visible interaction studies.

Application 2: Dynamics of UGC Platforms

  • Examination of the relationship between blogging and microblogging on Sina, identifying a weak competition and cooperative dynamics.

Application 3: Mobile Phone Usage Fragmentation

  • Analyzed user sessions on mobile devices using mobile phone logs, concluding weak evidence supporting fragmentation hypotheses.

Content Mining

What is Content Mining?

  • Content mining is seen as an evolution from content analysis, aimed at analyzing vast datasets of user-generated and traditional media content.

Methodological Differences from Content Analysis
Differences in Data Sourcing, Structure, and Scale

  • Heterogeneous data sources, unstructured content, and exponentially larger datasets conducive to dynamic analysis.

Methodological Strengths of Content Analysis in Content Mining

  • Traditional metrics help ensure rigorous methodological practices, even as content mining scales and automates analyses (e.g., reliability checks using Krippendorff’s alpha).

Techniques Utilized in Content Mining
Traditional Methods and Advances in NLP

  • Early dictionary approaches, transitioning to machine learning techniques and transformer models capable of nuanced analysis of social media content.

Application of Content Mining
Public Opinion Analysis

  • Empowered researchers to gauge real-time public sentiment across social media platforms, revealing shifts in public discourse and societal response to events.

Information Production and Flow

  • Social media's unique structure allows researchers to study information dynamics, highlighting censorship, misinformation, and diverse engagement patterns.

Challenges Specific to Content Mining in the Chinese Context

  • Overarching issues like algorithm contamination, balkanization of platforms, and ethical concerns around user data reflect the complexities of conducting research in China.

Computational Experiments

Definition

  • Computational experiments utilize online platforms for creating and conducting experiments, representing advancements in experimental design in the communication sciences.

Advantages Over Traditional Methods

  • Computational experiments enhance external validity and respondent diversity, allowing real-time, unobtrusive data measurement.

How to Conduct Computational Experiments

  • Methods resemble user analytics and content mining, enriched by collaborations with social media platforms to optimize operational effectiveness.

Empirical Cases

  • Notable studies illustrate varied applications of computational experiments, revealing insights into public attitudes and communication effects during high-stakes events like elections.

Conclusion and Future Directions in CCR

Conclusion

  • CCR embodies a fusion of traditional communicative inquiries with novel computational methods, providing robust frameworks for addressing long-standing communication debates. Future advancements in technology may further redefine the role of computational techniques in evolving research methodologies, influencing both communication scholars and societal dialogues.

Challenges Ahead

  • Address ethical considerations and ensure regulatory compliance, particularly in politically sensitive contexts like China, while continuing to leverage digital methodologies for enhanced communication research.