Automatic multi-class classification of patent documents
| Autor: | Chien-Nan Lin, 林建男 |
|---|---|
| Rok vydání: | 2010 |
| Druh dokumentu: | 學位論文 ; thesis |
| Popis: | 98 In recent years, intellectual property has attracted much attention from enterprises. To enhance the corporate competitiveness, the enterprises should constantly develop innovative products. However, patent analysis is a critical job for the Research and Development Department, then classification of patent document is one of the main tasks. Most patent search is still conducted manually due to the lack of reliable computerized tools. In addition, the lack of objective definition influences the quality of the patent search results. Therefore, the purpose of this paper is to propose a system to maintain a high level of consistent classification, assist the patent analysis and shorten the development time. In this paper, an automatic classification system is developed for patent documents classification. We apply the techniques of Support Vector Machine (SVM) and Vector Space Model (VSM) for classification. Two types of input data sets, the input features selected by the Sequential Forward Selection (SFS) and the input features without been selected by the SFS, are applied for performance comparison. The performance comparison is conducted under the combination of two types of input features and two types of classifiers. The Precision、Recall and F-measure are applied as the indices of the performance comparison. Also the statistical significance test is examined by the McNemar’s test. Experimental results show that the data set processed by the SFS and classifiers have out-performed the input features without been selected by the SFS in feature selection problem. The features selected by the SFS with the DAG-SVM classifiers provide the best classification performance (F-measure = 0.798). The second to the best are the for OAO-SVM classifier, the F-measure of SFS with features selected by the SFS with the OAO-SVM classifiers (F-measure = 0.774). |
| Databáze: | Networked Digital Library of Theses & Dissertations |
| Externí odkaz: |
|