Throughout school and university, we are taught about the importance of graphs to map information. We've been using them for more than 200 years now Amazingly, I can't really find any recent changes to these old-school grafts despite our massive advances in technology.
What I'm working on at the moment is finding a new kind of graph or illustration to integrate all of the information I have. I'd need to show different dates, different people, different events while subcategorizing this information into different geographical areas, different eras and different genres. There might even be more categories that need to be represented.
I've been looking into Emma Willard's contribution to Women's Education as pictured in the Temple of Time above for inspiration, it'd either have to be a structure like that, or a network resembling a spider's web.
Given the advances in technology, I'm surprised that we haven't seen more interactive and exciting ways to map out information. Imagine building an information structure like Willard's temple, but being able to open doors to different rooms or take a stairway to a different level.
Does anyone have any other ideas for how to approach illustrating the abovementioned information in a coherent manner?
I think a 3D Cube could be a good way to approach this. Every face, edge and vertice could be used to show different things and a specific point's information can be indicated by its position in relation to these points.
I also think this would be the theoretical basis of building a structure of information.
Based on my general comment below that most humans can understand the information provided in 2D or 3D and not in multiple dimensions, we could use embedded 3D cubes to visualize complex datasets. I am not completely sure about this; so a person with related experience can help me out.
Data in multiple dimensions can be structured in such a way that only 3 dimensions exist at each level. For example, if there are 8 dimensions to your data, the closest ones can be combined and a maximum of 3 different groups can be formed. This can be done using the phylogenetic tools that construct phylogenetic trees based on the similarity between multiple organisms. These 3 groups can then be visualized (along with their interactions with each other) using a 3D graph similar to the one created to visualize principal components.
Here, your 8 initial dimensions will now be reduced to 3 dimensions - two of them representing 3 original dimensions and one of them representing 2 original dimensions (since you had a total of 8 original dimensions). This information will be displayed below the axes labels of the cube. You can then click on one of the axes that you want to explore further and enter another 3D cube (second level) displaying that information. Here, the information would not be condensed and the original 3 dimensions will be displayed. Going from one cube to the other is like going a level deeper into the data.
So the approximation (merging of two or more dimensions into one) decreases as you go deeper into the dataset and it increases as you come out. At each level (3D cube), you are provided the information that is the best representation of data on all the levels that fall under that 3D cube.
Would it make things more clear or more complicated? (just a thought)
JuranApr 09, 2021
What I want to say is, would a new way of displaying information, be more understandable for an average user?
Picture having few simple 2d graphs with mass, width, length, and height of dogs. On one graph you have mass vs width, on the other mass vs length, and so on. They are pretty understandable, but limited, I agree.
Now combine mass, length, and height into one single graph with 3 dimensions. The things become interesting, but you need more time to understand them and extract information. If we combine 4 or more elements, things become a bit messy, but new clues could emerge after we analyze them.
The point is, it gives you more info and helps you discover new connections (a plus), but takes more time and computational power to be generated, is generally less understandable (requires more time to comprehend) and could result in overseeing some simple but valuable relations between variables.
I am sure it depends highly on the field it is applied to, but should we maybe also focus on better analysis of the existing graphs? Some software that would help us extract connections and clues between several separate graphs and variables they share?