Diverse samples

[BabakHemmatian]

• Random comment samples
• Probabilities of all top topics for each sample
• Different categories of top topics based on the values/consequences distinction

[sabjoslo]

@BabakHemmatian, is this still relevant?

[BabakHemmatian]

@sabjoslo

• Probably not. Steve pointed out that given how variable what people talk about can be, Just randomly sampling from the comments might not give us clearly value-based or consequence-based comments and it would be best to have SOME basis for choosing the comments. But it depends on how our top topic comment samples turn out.
• It would still be helpful to get probabilities of all top topics for each sample. It would tell us how entangled our various topics are
• Whether the last one is relevant depends on how neatly our top topics can be classified as based on values/consequences