Analyzing Network Effects on a Fanfiction Community

Andrés Carvallo, Denis Parra, Eduardo Graells-GarridoarXiv 2019 · OpenReview

TL;DR

This paper investigates what makes fanfiction authors popular on the fanfiction.net platform by analyzing user profiles and social network metrics. They found that while factors like the number of stories written are important, network features like closeness centrality significantly improve the prediction of author popularity, suggesting that community connections play a crucial role.

What Problem Does This Paper Address?

Online communities are a significant part of the modern internet, and understanding the dynamics within them is crucial. Fanfiction.net, a large community where users adapt and create stories based on existing media, presents a unique environment for studying social interactions and popularity. While previous research on fanfiction.net has focused on general data analysis, this paper aims to fill a gap by specifically exploring the role of social network analysis in understanding author popularity. The core question is: can social network features help predict which authors become more popular than others on this platform?

Methods

The authors employed a two-pronged approach using logistic regression models to predict author popularity, defined by having an in-degree greater than one (meaning at least one other user has them as a favorite author).

  1. Model 1 (Profile-based): This model used features derived directly from user profiles, including the number of written stories, favorite stories, biography word count, number of favorite authors, and the time since joining the platform.
  2. Model 2 (Network-enhanced): This model built upon Model 1 by incorporating social network metrics. These metrics included betweenness centrality, closeness centrality, PageRank, and clustering coefficient, all calculated from the "favorite author" relationships within the community.

The network was constructed by considering users who had at least one favorite author, with a relationship established if one user listed another as a favorite.

Datasets

The study utilized a dataset of 83,600 user profiles scraped from fanfiction.net between May 20th and June 5th, 2017. This sample was derived from a larger pool of approximately 9.23 million stories. The data extracted included user join dates, biographies, written stories, favorite stories, favorite authors, and community affiliations. For the network analysis, a subset of 1,734 users who had at least one favorite author was used, resulting in 2,766 interactions (edges).

Key Insights and Ideas

  • Time is not your friend (for popularity): Surprisingly, the longer a user has been on fanfiction.net, the less likely they are to gain followers. This suggests that newer users might be more active in building connections, or that older users become more passive over time.
  • Biography length is a double-edged sword: While a longer biography might seem like a way to connect with readers, the study found it has a negative effect on author popularity. This could imply that overly detailed or less engaging biographies don't necessarily translate to more followers.
  • Network structure matters more than you might think: The inclusion of network metrics significantly improved the predictive power of the models. This highlights that how authors are connected within the community is a strong indicator of their potential popularity.
  • Centrality is key: Closeness centrality, in particular, emerged as a significant predictor of author popularity. This suggests that authors who are "closer" to others in the network (i.e., can reach many others in fewer steps) are more likely to be followed.
  • The "Lord of the Land of Fire" phenomenon: The network analysis revealed a highly centralized structure with one dominant node, "Lord of the Land of Fire," concentrating a vast number of in-degree relations. This indicates that popularity can be highly skewed on the platform.

Key Results

  • The model incorporating network metrics (Model 2) showed a significantly higher fitness (lower AIC and deviance, higher Pseudo R-squared of 0.34) compared to the profile-only model (Model 1, Pseudo R-squared of 0.1188).
  • Stories written was the most significant factor across both models, positively impacting the probability of having followers.
  • Time spent on fanfiction had a negative impact on the probability of gaining followers in both models.
  • Closeness centrality was a significant predictor of author popularity in Model 2.
  • Betweenness centrality had an inverse relationship with the probability of having followers, meaning higher betweenness centrality was associated with fewer followers.
  • Clustering coefficient and PageRank were not found to be significant predictors in the network-enhanced model.

Limitations and Future Work

The authors acknowledge that the fanfiction.net network is not very dense, with many users preferring to read rather than write, and many having empty profiles. This lack of density might influence the network metrics.

For future work, they suggest exploring:

  • The impact of story reviews on author popularity.
  • Incorporating other factors like story language, genre, or the writer's country of origin.
  • Further investigation into how network effects influence user behavior and community dynamics.

Summary updated

Abstract

Since the early days of the Web 2.0, online communities have been growing quickly and have become important part of life for large number of people. In one of these communities, fanfiction.net, users can read and write stories which are adapted, recreated and modified from original famous books, tv series, movies, among others. By following stories and their authors, the fanfiction community creates a social network. Previous research on online communities has shown how features of the social network can help explain the behavior of the community, so we are interested in studying fanfiction's social network as well as its influence in aspects of the community. In particular, in this article we describe several properties of the members of the community, and we also try to discover which factors explain the popularity of the authors. We discover that time since joining fanfiction and the size of the authors' biography, has a negative effect on the authors popularity. Moreover, we show that the users' network metrics help to explain better authors' popularity.
BibTeX
@misc{carvallo2019analyzing,
  author = {Andrés Carvallo and Denis Parra and Eduardo Graells-Garrido},
  title = {{Analyzing Network Effects on a Fanfiction Community}},
  year = {2019},
  eprint = {1909.02886},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/1909.02886},
}
Figure