The original idea behind the representation of aspects was to have a data tag in the content description neuron named "Aspect", that named by which aspect the content described a concept. However, this design seemed unattractive when considering the multidimensional aspect filters that we wanted to realize. The realization that many of those other dimensions (such as school level) probably would be represented in the meta-data, led us to make the decision to put the "Aspect" tag amongst meta-data as well. This way aspect filtering becomes meta-data filtering, which in turn actually allows any aspect filtering system to work as a general neuron filtering and searching system, thanks to the fact that content descriptions are neurons like any others.
The hierarchy represented by the filter nodes is essentially a directed acyclic graph. This graph could often resemble a tree-structure, but not always, as the branches of the tree are allowed to grow together (even for sound filters). For very simple filters with a single level of refinements, a list presentation is reasonable. With two levels and the same level-two refinements in all level-one refinements, a matrix presentation is the most obvious.
In general, a menu with recursive sub-menus can examine any sound filter "locally", meaning that you cannot be sure to get a complete overview of the filter in this way.
More fitted to get an overview would be to present the filter in a diagram, in such a way that each filter node is represented by a box, connected to the rest of the filter nodes using arrows. In this way, each filter is a diagram showing an acyclic directed graph, where each filter node in addition is connected to its sub-filter, if any. These sub-filters could be incorporated into the diagram and connected to the filter node by a different type of association. Alternatively, each filter node can be connected to new diagram showing these subfilters.
It should be obvious that filter neurons are perfectly adjusted to be presented in concept maps. Each such diagram could be arranged in different ways, naturally: as a star, tree etc. This concept map would probably be added as content for the filter. Note that some refinements to the top filter neuron will have their own maps if they are seen as top level filter neurons for their branch.
This far, only meta data filtering has been implemented. To extend the filter usability outside of content filtering, one should consider other types of criteria when the packets are neurons or even concept maps, such as axon information or contained neurons.