Skip to main content
. 2022 Feb 21;2021:1264–1273.

Table 6.

Data overlapping among datasets in Group 2.

Scale Twitter COVID-19 stream Twitter multi-crawler search Twitter COVID-19 stream vs. Twitter multi-crawler search Twitter COVID-19 stream Twitter multi-crawler search Twitter COVID-19 stream vs. Twitter multi-crawler search
Experiment Comparison of all tweets Comparison of tweets with the top 10 tweets keywords
Minutea 62.8% (61.9%, 63.8%) 76.0% (74.2%, 77.8%) 51.8% (50.4%, 53.2%) 78.0% (74.9%, 81.1%) 98.4% (97.7%, 99.1%) 77.3% (74.2%, 80.4%)
Minuteb 73.7% (72.8%,74.6%) 78.9% (77.4%, 80.5%) 52.6% (58.8%, 61.8%) 75.8% (74.0%, 77.6%) 84.5% (83.8%, 85.2%) 60.3% (58.8%, 61.8%)
Hourb 67.1% (51.3%, 82.7%) 82.3% (68.5%, 95.5%) 50.5% (30.1%, 62.9%) 74.52% (73.5%, 75.5%) 84.1% (83.7%, 84.5%) 58.6% (57.8%, 59.4%)
Dayb 69.2% (57.3%, 80.8%) 83.1% (73.4%, 92.6%) 52.1% (43.4, 60.6%) 74.42% (74.2%, 74.7%) 84.5% (84.3%, 84.9%) 58.9% (58.2%, 59.7%)
Weekb 68.3% (57.3%, 80.8%) 83.3% (69.3%, 96.7%) 50.3% (42.0%, 58.0%) 74.5% (74.2%, 74.8%) 84.3% (84.0%, 84.6%) 58.8% (58.4%, 59.2%)
a

The denominator is the “Twitter full archive” dataset.

b

The denominator is the union of “Twitter COVID-19 stream” and “Twitter multi-crawler search” datasets.