| 摘要 |
A system for information item categorization in which each non-leaf node in a hierarchical organization of nodes represents a category, and each leaf node represents an information item. A number P is associated with each node. For non-leaf nodes, the associated number P is selected from a subset of relatively prime elements chosen from an appropriate Unique Factorization Domain (UFD), such as a set of relatively prime numbers which are a subset of the familiar set of integers. For leaf nodes, P is set to 1. A number M is also associated with each node. For each root node, M is set to the value of P for that node. For each non-root node, M is set to the product of the C's of all parent nodes of the node and the P of that node. As a result, the value of M for any node contains as factors the Ps of all its ancestor nodes in the tree or acyclic graph making up the node hierarchy. Queries to find items located under a given node N are accomplished in the disclosed system by identifying those leaf nodes having an associated value M that is completely divisible by the value of P associated with N. A node N is re-parented from parent node P 1 to parent node P 2 by removing the M factors of its old parents by dividing them out of the value of M for the node N, and adding the M factors of the new parents by multiplying the M value of the node N with these factors.
|
