VERY-LARGE-SCALE AUTOMATIC CATEGORIZER FOR WEB CONTENT

摘要

A kind of method and apparatus are provided in figure and the ultra-large hierarchy classification tree of audio constituent instruments for effectively classification and classification data object, such as e-text. According to one embodiment, the selection of the object hierarchies of the first node in multiple nodes. Previous classified data object (202) corresponds to the object hierarchies of the first node of selection and the node rendezvous of any associated child node to form content class data object (206). Similarly, the brother of node selected node of the corresponding object of data and any associated child node brother of node are then assembled to form the data object of non-content grade (206). Then (207) of feature extraction are opposed from the data of each content class data object and non-content classification convenient for the previous classified data object (202) of characterization (209).