Comment analysis

Detailed Comment Analysis

A deeper dive into the semantic analyses of the 9,910 comments, including thematic analyses, post perspectives, and organizations or names mentioned in the posts.

Responses Discuss Specific Social Media Platforms, Politics, and Social Life

All 9,910 top comments, pooled together regardless of parent post, then embedded and clustered. Due to the shorter length of the comments, there was a higher number of clusters and more noise. Comments most frequently discuss Reddit, Youtube, politics, and social life.

One cluster below is flagged boilerplate: a grab-bag of short reaction comments ("true", "aw", "sad", "bye") — shared phrasing, not a real topic.

Redditors also Respond Focusing on Themselves

Similar to the post analysis, the distribution of posts in first, second, and third-person point of view was computed to determine whether posts are typically about themselves or other people. From the analysis, first-person language is seen to still be most commonly used, although the number is lower (39.1%) compared to posts (63%).

Who are Redditors Mentioning in the Comments?

The same named-entity recognition pipeline was conducted was over every comment. 2,191 comments name at least one organization and 542 comments name at least one person. This is much higher than posts, likely due to the larger number of comments overall.

Comments have a Balanced Mix of Sentiment Across Themes

Sentiment crossed against the comment clusters above, sorted by share negative. While many themes still have a high number of negative comments, there are also many themes with a more balanced mix of sentiment, and even some themes with a majority of neutral or positive comments.

0%50%100%
Negative Neutral Positive