Influential STM Tweets by Topic (k = 10)

Influential Tweets per Topic

This document presents, for each STM topic (\(k = 10\)), the tweets that are both:

• Strongly associated with the topic: we require a topic probability \(\gamma \ge 0.5\) for that topic.
• Highly influential in terms of engagement: we compute an engagement score \[ \text{engagement\_score} = \text{retweet\_count} + \text{like\_count} \] using the tweet-level metadata from the social network analysis (uniques_final_chatgpt).

For each topic, we select the top 20 tweets by engagement score (breaking ties by higher \(\gamma\)), and combine them into a single interactive table.

This gives us a set of influential exemplars: tweets that not only best express each topic linguistically, but also received substantial attention in the retweet/like economy.

stm_influential_dt

Interpretation and Use

• Analytical role: These influential exemplars complement the earlier table of top-gamma tweets by focusing on visibility and diffusion rather than only linguistic purity. They highlight which framings of each topic actually shaped the wider conversation.
• Reading strategy: For a given topic, filter by the topic number, then sort within that topic by engagement_score, retweet_count, or user_followers_count to inspect the tweets and accounts that carried the topic most effectively through the network.
• Methodological note: Because engagement is unevenly distributed across users (e.g., large accounts vs. ordinary users), this table should be read alongside the full topic distribution and network analysis to avoid over-attributing the topic’s meaning to a small number of highly visible tweets.