[Paper Review] Community embeddings reveal large-scale cultural organization of online platforms

Isaac Waller, Ashton Anderson|arXiv (Cornell University)|Oct 2, 2020

Social Media and Politics27 references4 citations

TL;DR

This paper introduces a neural embedding method to map online communities along cultural dimensions using aggregate behavior in 4.8 billion Reddit comments across 10,000 communities over 14 years. It reveals that platform-wide polarization, especially post-2016 U.S. election, is driven by new political users rather than individual radicalization, and that online cultural structures are more complex than offline analogues.

ABSTRACT

Optimism about the Internet's potential to bring the world together has been tempered by concerns about its role in inflaming the 'culture wars'. Via mass selection into like-minded groups, online society may be becoming more fragmented and polarized, particularly with respect to partisan differences. However, our ability to measure the cultural makeup of online communities, and in turn understand the cultural structure of online platforms, is limited by the pseudonymous, unstructured, and large-scale nature of digital discussion. Here we develop a neural embedding methodology to quantify the positioning of online communities along cultural dimensions by leveraging large-scale patterns of aggregate behaviour. Applying our methodology to 4.8B Reddit comments made in 10K communities over 14 years, we find that the macro-scale community structure is organized along cultural lines, and that relationships between online cultural concepts are more complex than simply reflecting their offline analogues. Examining political content, we show Reddit underwent a significant polarization event around the 2016 U.S. presidential election, and remained highly polarized for years afterward. Contrary to conventional wisdom, however, instances of individual users becoming more polarized over time are rare; the majority of platform-level polarization is driven by the arrival of new and newly political users. Our methodology is broadly applicable to the study of online culture, and our findings have implications for the design of online platforms, understanding the cultural contexts of online content, and quantifying cultural shifts in online behaviour.

Motivation & Objective

• To address the challenge of measuring large-scale cultural organization in pseudonymous, unstructured online communities.
• To understand how online cultural structures differ from their offline counterparts.
• To quantify shifts in cultural alignment on online platforms over time.
• To investigate the drivers of polarization in online discourse, particularly around major political events.

Proposed method

• The study employs a neural embedding model trained on aggregate comment patterns to represent communities in a continuous cultural space.
• Embeddings are learned from co-occurrence statistics of community-level language use across 4.8 billion Reddit comments.
• Cultural dimensions are inferred from the geometric structure of the embedding space, capturing latent cultural relationships.
• The method enables tracking of community positioning and cultural drift over 14 years of Reddit activity.
• Temporal analysis is applied to detect structural shifts, particularly around the 2016 U.S. presidential election.
• The approach is validated by examining political content and comparing cultural relationships to real-world analogues.

Experimental results

Research questions

• RQ1 How is the macro-scale cultural structure of online platforms organized along cultural dimensions?
• RQ2 To what extent do online cultural relationships reflect or diverge from their offline counterparts?
• RQ3 What role do new users play in driving platform-wide polarization over time?
• RQ4 How did the 2016 U.S. presidential election affect the cultural organization of Reddit communities?
• RQ5 Are individual users primarily responsible for increased polarization, or is it driven by demographic shifts in user composition?

Key findings

• The macro-scale structure of Reddit communities is systematically organized along cultural dimensions, revealing a large-scale cultural organization.
• Cultural relationships among online concepts are more complex than simple reflections of their offline counterparts.
• Reddit experienced a significant polarization event around the 2016 U.S. presidential election, with sustained polarization for years afterward.
• Individual user polarization over time is rare; the majority of platform-level polarization stems from the influx of new, politically oriented users.
• The arrival of new users, particularly those with political leanings, is the primary driver of increased platform-wide polarization.
• The proposed embedding methodology effectively captures cultural dynamics and enables quantitative analysis of large-scale cultural shifts in online discourse.