Hierarchical categorization method and system with automatic local selection of classifiers

摘要: The present invention relates generally to the classification of items into categories, and more generally, automatic selection different classifiers at places within a hierarchy categories. An exemplary hierarchical categorization method uses hybrid technologies, with training-data based machine-learning preferably being used in those portions above dynamically defined boundary which adequate training data is available, a-priori rules not requiring any such below that boundary, thereby providing novel technology capable leveraging strengths its components. In particular, it enables use human-authored finely divided towards bottom involving relatively close decisions for practical create advance sufficient ensure accurate by known algorithms, while still facilitating eventual change-over machine learning algorithms as becomes available acceptable performance particular sub-portion hierarchy.