A Multilabel Text Classification Method with Personal Preference
碩士 === 國立成功大學 === 資訊管理研究所 === 96 === Automated text categorization has been widely used in many fields; it is the best solution to mass document management. Currently, most classification techniques have been applied to text categorization. However, these researches do not include personal preferenc...
| Main Authors: | , |
|---|---|
| Other Authors: | |
| Format: | Others |
| Language: | zh-TW |
| Published: |
2008
|
| Online Access: | http://ndltd.ncl.edu.tw/handle/57014951613059059156 |
| Summary: | 碩士 === 國立成功大學 === 資訊管理研究所 === 96 === Automated text categorization has been widely used in many fields; it is the best solution to mass document management. Currently, most classification techniques have been applied to text categorization. However, these researches do not include personal preference in their classification methods. Classification results depend on personal preference, hence different users may not label the same class to the identical document. This is usually ignored in text categorization so far. The purpose of this study is to find if personal preference will affect classification results, and to improve the classification effectiveness. We use back-propagation neural network (BPN) to build a preference-based text categorization model. The well-known Reuters-21578 collection is used to perform experiments. Experiment results show that the preference-based model is superior to the original one.
|
|---|