Table 1.
Performance of the five adopted classifiers for topic detection and sentiment classification.
Fig 1.
Share of VAR tweets and mean sentiment for each minute of Liverpool FC vs. Manchester United FC.
The grey shadowed boxes display the procedure described in statistical analysis.
Fig 2.
Sentiment distribution of all tweets (Total tweets), tweets related to the VAR (VAR tweets), and tweets not related to the VAR (Other tweets).
Negative tweets are denoted by red, neutral by orange, and positive by green color. The black line denotes the average sentiment score.
Fig 3.
Number of tweets related and not related to the VAR per vigintile.
The dashed line in dark grey displays the expected number of VAR tweets, the one in light grey illustrates the expected number of tweets not related to the VAR for each vigintile.
Fig 4.
Mean sentiment for the nine five-minute windows before and after VAR incidents.
The results of the paired t-test for the comparison of each window with the immediate window before the incident (MeanSentInterval_Before_1, grey bar with red border) are displayed in the respective bars.