The Twitter Analysis.

Seeing the high level of online activism and engagement on Twitter, we analyzed what Twitter users were talking about, categorizing sentiments about the Black Lives Matter movement.

Sentiments.

 

We categorized Twitter posts by sentiment around the Black Lives Matter movement from May 25, 2020 to June 25, 2020.

Negative tweets were highly prevalent, especially at the beginning stage of this protest life-cycle. More about how we define sentiments here.

10k Tweets were extracted from the hashtags “#blm” or “#blacklivesmatter” each day during the collection period. A sentiment analyzer was built based on the Flair Model, trained with data sets from SemEval. The accuracy of trained Flair was about 7-% on the test set. The average sentiments were calculated using the formula: (#tweets with positive sentiment - # tweets with negative sentiment)/#tweets. This indicated the "net sentiment ratio.”

Our analysis split tweets into three sentiment categories: positive, negative, and neutral. Positive tweets included terms that have more positive connotations; Negative tweets included more terms that have more negative connotations. Neutral tweets did not use particularly positively or negatively connotative language. Positive/Negative/Neutral very narrowly defines the emotional state ("sentiments") in which online users were discussing issues around BLM.

Most of the tweets were negative, especially in the first couple of days following George Floyd’s murder. Later on, as the movement transitioned into its peak stage, sentiments became more stable. Still, 60% of tweets were classified into the "negative" category. (The only major exception to this trend was seen on Juneteenth.)

 

Hashtag topics.

 

We visualized the rise and fall of trending topics around the Black Lives Matter movement from May 25, 2020 to June 25, 2020 based on trending Twitter hashtags.

Trending topics evolved from focusing on the individual tragedy around George Floyd’s death to focus more broadly on long-term actions (e.g. defund police).

10k Tweets were extracted from the hashtags “#blm” or “#blacklivesmatter” each day during the collection period. Hashtags similar to “#blacklivesmatter” was excluded from the collection, and the top 15 hashtags for each day were collecte and mapped into a certain category.

The visualization shows the rise and fall of topics, which can be separated into three distinct stages: the Growing Stage, Peak Stage, and Aging Stage:

Growing Stage: In the early growing stage, hashtags were focused around George Floyd’s murder, (e.g.  #georgefloyd, #icantbreathe, #justiceforgeorgefloyd). At the end of May, protest-related hashtags were emerging, which was followed by hashtags associated with specific activities, such as #blackouttuesday.

​Peak Stage: #blackouttuesday was at the peak stage of this protest life cycle. Around that time, Breonna Taylor's murder gained online traction, as well as criticism around the police system. Approaching the end of this stage, #defundpolice stepped into the top ten trending hashtags for the first time, along with a growing number of unrelated hashtags.

​Aging Stage: The stage is the most complex. Topics around other related events, and holidays (#juneteenth) came and went. Tangentially related topics (e.g. antifa) were drawing the public's attention. Nevertheless, #defundthepolice was stable within the list of top ten trending hashtags throughout the rest of June. At the end of June, #georgefloyd was no longer the top related hashtags to BLM, but #defundthepolice was.

Read on about Analysis by Topic