Communicative patterns on Twitter: An experiment in large-scale quantitative analysis
The project leverages the Twitter Firehose to examine distinct conversations around various issues including human rights, political communication, and development and humanitarian topics. It seeks to identify large-scale qualitative patterns using metrics such as the volume of posts over time, the top Twitter hashtags associated with, or triggered by the original conversation hashtag, the most mentioned Twitter authors within a conversation by the number of posts that mention that author, the top retweets, the most shared content and websites in the Twitter posts, the geographical distribution of conversations by country, and the most frequently used terms in a conversation.
Focus and queries
The project aims to contribute to research examining the dynamics of public discourse online, and to inform program development, implementation, and evaluation. Specific queries include:
- Volume and trend
Examine the volume trends in various conversation contexts. General queries include examining the conversation against the political and social backdrops, correlations between peaks and factors on the ground, and signs of artificial amplifications of content. Context-specific queries vary depending on the distinct conversation. Examples include finding if development and humanitarian issues receive steady or transient attention on Twitter and if campaign-driven peaks maintain momentum.
- Associated topics and alternative or counter narratives
Examine if a conversation snowballs to include multiple concurrences of thematically similar or dissimilar hashtags. Thematically related hashtags in a conversation indicate that users attempt to introduce similar topics or subtopics. On the other hand, the prevalence of counter-hashtags suggests that users seek to challenge the narratives. Irrelevant hashtags often seek to dilute or hijack the conversation.
Identify and measure salience of stakeholders in conjunction with the conversation by examining the mentioning behaviors of the participating Twitter accounts. Accounts use ‘@’ and a screen name in a tweet as a means of engagement.
- Content informing the conversation
The most popular content and websites shared by the users represent the type of content that informs the conversations, and that which resonates among the users.
- Geographic distribution of authors
Geographic metadata, when available, indicates the location where the tweet was authored. Identifying where most tweets are posted from and the countries that have the greatest influence on the conversation can help denote relationships between geographic propinquity, topical attention and users’ engagement.