Digital Text Categorization Based on Directional Term Structures

A rule learning classifier is capable of identifying potentially interesting patterns from training documents to establish classification rules. There have been a number of related studies for classifiers that utilize the characteristics of readable rules, such as association rule-based techniques and decision trees. These rule-based studies do not consider the structures of the terms in the documents. However, we believe that such structures may help reveal the core themes of the documents. Hence, this paper presents a new concept: Meaningful Inner Link Object (MILO). MILO classifies documents by finding the underlying directional link formed by the terms shared between different paragraphs. Moreover, a hierarchical classification structure, which considers the similarity between categories, is presented to improve the accuracy of the classification. The experimental results show that MILO is quite competitive when compared against other state-of-the-art classification techniques and may even surpass them in certain cases.

並列關鍵字

Automatic text categorization ； rule-based ； term distribution

國際替代計量

全文下載

主題瀏覽