Figure 1.
Keyword phrases used fuzzy matching to additionally capture close but not identical phrase matches, such as “not really science.”
Figure 2.
Examples of rejected candidate topics from 45 topic solution.
Ten most likely words for each topic listed to right of attempted topic label.
Figure 3.
Ten most likely words for each topic listed to right of assigned topic label. Asterisks indicate subjects previously identified in qualitative case studies of demarcation.
Figure 4.
Labeled substantive topics ranked by topic Dirichlet parameter.
The greater the topic Dirichlet parameter, the greater the proportion of the corpus assigned to the topic. Asterisks indicate subjects previously identified in qualitative case studies of demarcation.
Figure 5.
Labeled substantive topics ordered by rank_1 metric.
The rank_1 metric measures the number of times a topic is the primary topic in the documents in which it occurs, here expressed as a percentage. Asterisks indicate subjects previously identified in qualitative case studies of demarcation.
Figure 6.
Prominence of evolution in demarcation corpus over time.
Percentage of documents in each year with at least 15% evolution content.
Figure 7.
Prominence of race in demarcation corpus over time.
Percentage of documents in each year with at least 15% race content.
Figure 8.
Comparison of prominence of religion (dotted blue) and evolution (solid red) in demarcation corpus over time.
Percentage of documents in each year with at least 15% topic content.
Figure 9.
Prominence of presidential politics in demarcation corpus over time.
Percentage of documents in each year with at least 15% presidential politics content.