Difference between revisions of "Talk:Sugar Labs/SOM"

From Sugar Labs
Jump to navigation Jump to search
m
 
(2 intermediate revisions by the same user not shown)
Line 1: Line 1:
 
Fascinating! Any chance you can get a 'key' on the maps themselves so we can see what the 'colors' mean without referring to the main page? --[[User:Dennis Daniels|Dennis Daniels]] 14:48, 9 August 2009 (UTC)
 
Fascinating! Any chance you can get a 'key' on the maps themselves so we can see what the 'colors' mean without referring to the main page? --[[User:Dennis Daniels|Dennis Daniels]] 14:48, 9 August 2009 (UTC)
: Thanks for the feedback. I did try that in the past, the challenge is that the maps are wraparound (surface of a torus) so a legend can't go on an edge without loosing the ability to tile/scroll/pan about - I do that a lot here off-line and would like to provide that ability online at some point. The legend would then need to be placed in the map, but would need to move position from map to map so as not to obscure actual map detail. I was hoping (by using a geographic metaphor) that the map would be obvious enough to read without a legend, so any suggestions for simplification would much appreciated! There are really only 2 major variables the map is trying to display: 1) proximity of terms indicates their similarity, 2) topographic height indicates (a combined measure of) term frequency & strength of similarity. There are some other boundary markers going on, but that's not really needed for reading the maps (they helps me keep an eye on cluster quality); also there is a subtle detail in that the term font size is based purely on its occurrence frequency, so you can spot occasional large labels in low valleys (frequent usage but used with a low consistency, peoples names can often exhibit this especially if they are involved in broad/general discussion topics), and small labels on mountain ridges (low frequency of usage but used in a high consistent way).
+
: Thanks for the feedback. I did try that in the past, the challenge is that the maps are wraparound (surface of a torus) so a legend can't go on an edge without loosing the ability to tile/scroll/pan about - I do that a lot here off-line and would like to provide that ability online at some point. The legend would then need to be placed in the map, but would need to move position from map to map so as not to obscure actual map detail. I was hoping (by using a geographic metaphor) that the map would be obvious enough to read without a legend, so any suggestions for simplification would much appreciated! There are really only 2 major variables the map is trying to display: 1) proximity of terms indicates their similarity, 2) topographic height indicates (a combined measure of) term frequency & strength of similarity. There are some other boundary markers going on, but that's not really needed for reading the maps (they help me keep an eye on cluster quality); also there is a subtle detail in that the term font size is based purely on its occurrence frequency, so you can spot occasional large labels in low valleys (frequent usage but used with a low consistency, peoples names can often exhibit this especially if they are involved in broad/general discussion topics), and small labels on mountain ridges (low frequency of usage but used in a high consistent way). --[[User:Garycmartin|Garycmartin]] 14:53, 13 August 2009 (UTC)

Latest revision as of 10:56, 13 August 2009

Fascinating! Any chance you can get a 'key' on the maps themselves so we can see what the 'colors' mean without referring to the main page? --Dennis Daniels 14:48, 9 August 2009 (UTC)

Thanks for the feedback. I did try that in the past, the challenge is that the maps are wraparound (surface of a torus) so a legend can't go on an edge without loosing the ability to tile/scroll/pan about - I do that a lot here off-line and would like to provide that ability online at some point. The legend would then need to be placed in the map, but would need to move position from map to map so as not to obscure actual map detail. I was hoping (by using a geographic metaphor) that the map would be obvious enough to read without a legend, so any suggestions for simplification would much appreciated! There are really only 2 major variables the map is trying to display: 1) proximity of terms indicates their similarity, 2) topographic height indicates (a combined measure of) term frequency & strength of similarity. There are some other boundary markers going on, but that's not really needed for reading the maps (they help me keep an eye on cluster quality); also there is a subtle detail in that the term font size is based purely on its occurrence frequency, so you can spot occasional large labels in low valleys (frequent usage but used with a low consistency, peoples names can often exhibit this especially if they are involved in broad/general discussion topics), and small labels on mountain ridges (low frequency of usage but used in a high consistent way). --Garycmartin 14:53, 13 August 2009 (UTC)