I have a tree that represents a hierarchical ontology of computer science topics (such as AI, data mining, IR, etc). Each node is a topic, and its child nodes are its sub-topics. Leaf nodes are weighted based on their occurrence in a given document.
Is there a well-known algorithm or function to calculate the weight of inner nodes based on the weights of leaf nodes? Or is it totally based on the application to decide mathematical calculation of the weights?
In my application, the node's weights should be some sort of accumulation of its child nodes weights. Is there a better mathematical formula or function to do that than just summing up weights of child nodes?
I am not asking about traversal, but rather about weighting function.