This paper describes how natural language processing and ontologies are exploited for automatic text categorisation. The approach introduced is part of the MANENT system, an infrastructure for integrating, structuring and searching Digital Libraries. The procedure of structural information extraction, and of the automatic classification of the records according to natural language understanding and theWordNet Domains taxonomy is discussed. A comparison between two versions of the classification algorithm is conducted and the improvements of the new approach are articulated. In particular, using semantic connections between words refines the classification results while reducing misclassification to non classification. © 2011 IEEE
When you doubt, abstain: From misclassification to epoché in automatic text categorisation
LOCORO, ANGELA;
2011-01-01
Abstract
This paper describes how natural language processing and ontologies are exploited for automatic text categorisation. The approach introduced is part of the MANENT system, an infrastructure for integrating, structuring and searching Digital Libraries. The procedure of structural information extraction, and of the automatic classification of the records according to natural language understanding and theWordNet Domains taxonomy is discussed. A comparison between two versions of the classification algorithm is conducted and the improvements of the new approach are articulated. In particular, using semantic connections between words refines the classification results while reducing misclassification to non classification. © 2011 IEEEI documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
