Skip to main content
. 2011 Aug 23;5:72. doi: 10.3389/fnhum.2011.00072

Figure 5.

Figure 5

Average classification accuracy across models using 10 to 100 topics, for each of 9 subjects (chance level is 0.5); the accuracy is broken down into classification of concept pairs where concepts are in different categories (“Between”) and pairs where the category is the same (“Within”). Error bars are across numbers of topics, chance level is 0.5.