| With the rapid development of technology and society, academia tends to beinformationization. The Internet retrieve becomes an important way to obtain relevantinformation for a vast number of scholars. Therefore, how to find the scientific papers quicklyand accurately in the electronic knowledge base has become a research focus for manyscholars, and increasingly attention by them. Text classification plays an important role inmany information retrieval systems and text mining, it can improve the retrieval performance,provide browsing and navigation mechanisms and discover similar text opportunely.According to the number of class label, text classification can be divided into:single-label classification and multi-label classification. In practical applications, multi-labelclassification is quite common. Multi-label classification of the current study focused onmulti-label classification of feature selection and classification algorithm. Already, theperformance of the existing multi-label feature selection algorithm is difficult to satisfactorily,and some are high time complexity, and some have little impact on the classificationperformance. In recent years, the field of computers widely used ontology the concept as ameans of knowledge organization and knowledge representation, in theory, has manyadvantages and potential functions. As the ontology is not just a collection of concepts, butalso embodies the concept of inter-linkages, therefore, the ontology and text classificationcombining with research significance.Recently, there are more and more Chinese science papers appearing on the Internet inthe form of electronic text, and they are concerned by majority of scholars. But research onscience paper classification is very little. Some research is mainly solving the single-labelclassification problem, they do not realize that one paper may belong to several predefinedtopics; and this will affect precision of classification. |
PDF Full Text Request