This is a follow up to our discussion about meta-data organization and structure. In our previous post we discussed the need to move from a tree structure to a graph (or digraph) structure. We also identified two key types of relationships between nodes, the "is" relationship and the "has" relationship. In one of our brainstorming sessions, Federico Prada, our Research Director, suggested using a multidementional graph to represent the meta-data, so we are starting to explore this idea.
Wikilife's meta-data is basically a set of life variables with context. For example, a person may want to track her nutrition, so the food is the life variable that she will be tracking, and things like portion size, location, company, time of the day, etc. are part of the context. Some of these context items are described by reduced list of options. Other context items (i.e. places, or body parts) have a complex domain that need to be reprsented by a more complex structure.
Taking into consideration Federico's idea, we are now exploring representing our meta-data as a one big graph of live variables (LVG) and several other graphs for the context. In this approach nodes from the life LVG are connected to nodes from the context graphs. So for example the node "headache" from the LVG is connected to the node "head" from the body parts graph (BPG).