Activity On Vertex 네트워크를 이용한 계층구조의 문서자동분류 기법
(An) Automatic Document Classification Technique of Hierarchical Structure Using Activity on Vertex Network
Activity On Vertex 네트워크 계층구조 문서자동분류;
- 원문 URL
As the number of electronic documents is increasing rapidly, it is getting more difficult to fine the useful knowledge information and control them properly. In this thesis, it has been suggested an automatic document classification technique of hierarchical structure using the activity on vertex network. We extracted categories' feature sets from learning documents with Centroid vector method and calculated similarity between each categories' feature sets and learning documents in cosine coefficient. We checked the mis-classification in learning documents and extracted them to be frequently occurred in the high categories. Then we constructed the AOV network on the basis of these data and classified the test documents on basis of the AOV network. In result, it is better then the case of classifying it in hierarchical structure of tree-type. To verify the efficiency, we used the 230 academic papers published on computer and electronics conferences and extracted the indexes. Finally we verified the efficiency by comparing to the classification performance in hierarchical structure of tree-type and in that of AOV network.