Descriptor definitions: A descriptor definition is the combination of all the user-concepts that share a descriptor (Bob’s “tinny”, Maria’s “tinny” and Sue’s “tinny” combine to form the descriptor definition for “tinny”). This table provides overview summary statistics about descriptors taught to SocialEQ by the public.
User concepts: A user-concept is an equalization curve that embodies a descriptor, as taught to the system by a single user (e.g. Maria’s concept for “tinny”). This table contains all the individual user concepts, encoded as 40-band equalization settings.
Descriptor definition distance table: We define a distance measure between descriptors here. This table contains all pair-wise distances between descriptive terms and lets us answer questions such as “Is ‘warm’ closer to ‘broad’ than it is to ‘tinny’?”.
Hierarchical clustering: We performed agglomerative hierarchical clustering of all descriptor definitions by using the "group average" algorithm and Kullback-Leibler divergence and plotted the resulting dendrogram here. See this article for more detail. download figure
2D Equalization Descriptor Map: We placed the descriptor definitions in a two-dimensional space by using metric multidimentional scaling. In this figure, each descriptor definition with at least 2 user-concepts is scaled by its agreement score so that well defined descriptors are larger. download figure (also seen below:)