Sentiment and hate content analysis of CrowdTangle datasets
Input dataframe -> Original Facebook data scraped from CrowdTangle
Output dataframe -> Input dataframe + appended sentiment scores predicted by the model
Input dataframe -> Output dataframe from Step 1
Output dataframe -> Input dataframe + appended hate speech scores predicted by the model
Input data -> Output dataframe from Step 2
Outputs -> Inline plots and stats.
- Step 4. Evaluate the performance of the hate speech prediction model on human annotated dataset
-- Step 4.1. Randomly sample representative dataset(s) for human annotations
-- Step 4.2. Cross-validate the annotated labels with model predictions