A sentiment analysis of the Black Lives Matter movement using Twitter
Bibliographic record
Abstract
As more attention is brought to the issue of racial injustice, public sentiments and opinions on racial issues are increasingly important to track. At the same time, recent progress in machine learning and natural language processing methods, coupled with the growing amount of available data for training and analysis, allows researchers to extract sentiments from text data at large scales. We applied a natural language processing framework to study public sentiment surrounding the Black Lives Matter (BLM) movement. Specifically, we used a state-of-the-art BERT model fine-tuned for Twitter sentiment classification to predict the sentiment from approximately 1 million tweets from July 2013 to March 2021 related to BLM. The BERT model was trained on the Sentiment 140 dataset on which it obtained an AUC of 0.97 on the training data and 0.94 on testing data, outperforming other machine learning models. We found that retweet frequency and word count frequency were able to illustrate important themes in the BLM movement as well as indicate events of significant importance to the movement. Additionally, sentiment analysis revealed which of these themes and events were associated with positive public sentiment, such as social justice, and which were associated with negative sentiment, such as police brutality. Our analyses can also be applied to better understand other social and political movements to aid related research and activism.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.001 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.002 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.001 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.001 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".