How do we measure "where" a category is represented?
How do we measure "where" a category is represented?
Ultimate Plan:
don't include words from target category
Note, I am not claiming this is sufficient, but that it is necessary
cat_101
Represents Category
cat_101
Does NOT Represent Category
This metric tells us how "distinct" each word is from the other words
Note: For each subject, for each category, we get a "mean percentile" map
1) Take presentation
2) Get distance to all other presentations
3) Take average rank of same word
4) Average words by category
This metric tells us how "distinct" each word is from the other words
Animal v AO
Animal v AO
Human Trait v AO
Quantity v AO
Toggle
(Average Distance between Blue and Orange Dots)
(Average Distance Between Blue Dots)
S =
Between group variance
Within group variance
Where are DSM's for a category robust at a group level
Instead of each point bein a presentation of a word to a subject, each point is a subject (with all 6 presentations averaged in that subject)
Toggle
CAT Noise
Toggle
SOE Noise