Text-based automatic content classification and grouping

摘要

A closed-caption [101], is passed to a natural language analysis tool [102]. Noun phrases and proper nouns in closed-captions are extracted and saved in a file [103]. The noun phrase file is passed to a word-code translation tool [104] and each different word is assigned a unique code from a dictionary. The output [105] of the word-code translation tool [104] provides source data for story classification [110] and grouping [114]. For story classification [110], training [107] and testing [111] examples are generated by another tool [106]. A story classification knowledge network [109] is generated from training examples [107] input to the training module and modified thereafter for each new story. Class prediction [112] and knowledge base modification can be realized interactively on a news organizer platform. Relevant story grouping [114] takes a story location [105] and corresponding story grouping files [113] and determines a group [115] for the new story.