Making Sense of Post-match Fan Behaviors in the Online Football Communities

Title of the Paper

Making Sense of Post-match Fan Behaviors in the Online Football Communities

Paper Information

• Subject Area: Behavioral Analysis in HCI and Social Networks
• Keywords: Online Communities, Fan Behavior, Professional Sports, Topic Modeling, English Premier League, Football, Social Interaction

Research Background and Issues

• Identified Problems or Challenges

  • Existing research has not thoroughly explored how offline football match events influence the sentiment and discussion content within online communities.
  • Negative comments, toxic behavior, and polarization within communities pose threats to peaceful interactions on platforms and among users.
  • There is a lack of comprehensive understanding of the dynamics of online sports communities, particularly the detailed connection between offline and online interactions.

• Significance of the Research

  • Online sports communities serve as a pioneering "sandbox" for studying how offline events impact online behaviors.
  • Investigating emotional shifts and community feedback contributes to understanding the social dynamics of online communities.

• Motivation and Related Work

  • Intense match atmospheres often lead to expressions of extreme emotions in online discussions; studying these characteristics can aid in designing safer community ecosystems.
  • Existing research primarily focuses on user activity and fan loyalty, with limited analysis of comments and discussion content.

Proposed Solutions

• Proposed Methods and Solutions

  • Introduced the "gap score" metric to quantify the impact of offline match events on community behavior.
  • Leveraged Natural Language Processing (NLP) techniques for sentiment analysis and structured topic modeling.
  • Conducted quantitative analysis using comment data from the English Premier League Reddit community.

• Innovations

  • First to propose the "gap score" metric based on sentiment and expectation gaps to quantify the impact of offline events on online discussions.
  • Utilized a large-scale dataset (over 177,000 posts and 3.7 million comments) for detailed analysis.
  • Employed Structured Topic Modeling (STM) to uncover discussion content characteristics and analyze sentiment changes.

• Implementation Steps and Key Techniques

  1. Collected and preprocessed post and comment data from Reddit communities.
  2. Conducted sentiment analysis on comments using the VADER tool to generate sentiment scores.
  3. Used STM to analyze the main topics of discussion and explored how discussion themes changed with match outcomes.
  4. Performed quantitative analysis of influencing factors and validated correlations using linear regression.

Research Findings

• Specific Findings

  • Sentiment analysis results indicate that team match outcomes significantly affect community comment sentiment, with more positive sentiment following victories and more negative sentiment after losses.
  • Discussion content primarily falls into four categories: match process, season performance, lineup discussions, and personnel adjustments.
  • The "gap score" is significantly positively correlated with fan sentiment, with team performance exceeding fan expectations leading to higher positive sentiment.

• Comparison with Existing Solutions

  • Compared to other studies, this research not only examines user activity but also thoroughly evaluates sentiment flows and their association with discussion topics.
  • Provides broader data support and quantitative analysis.

• Experimental or Evaluation Results

  • Users tend to upvote comments with strong emotions, while neutral or moderate content receives less attention.
  • Polite comments do not receive significantly more upvotes than other comments, indicating that community feedback focuses more on emotional expression than linguistic form.

• Limitations and Future Directions

  • The data is limited to the English Premier League, and the geographic and event scope constraints may affect the generalizability of the research.
  • Extending the study to other sports and community types, such as F1 or national team matches, could provide more universal insights.
  • Incorporating non-text content such as videos and images into the analysis is a future direction.

Conclusion and Insights

• This paper demonstrates the significant impact of offline events on the sentiment and topic dynamics of online sports communities.
• Provides data support for community design, such as emotion regulation mechanisms and effective content moderation rules.
• Offers an effective analytical framework combining linear regression and topic modeling, contributing to future research and optimization of online social networks.