Weight of Features in Automatic Data Classification

• Main Authors: CHANG-MING CHEN, 陳昶旻
• Other Authors: Chung-Shyan Liu
• Format: Others
• Language: zh-TW
• Published: 2013
• Online Access: http://ndltd.ncl.edu.tw/handle/77191503658629811230

碩士 === 中原大學 === 資訊工程研究所 === 101 === It will be difficult for users to find needed documents if documents are not properly classified. Feature selection and screening in is an important step in automatic document classification. The documents in a Knowledge Management System are stemmed and then the weight of each term associated with a document is calculated using TF-IDF. WordNet was used to screen the relevant keywords from example recipe files to compose the feature vectors. Cross-Validation method was used to train the training model. The unclassified documents and then classified using k-nearest-neighbor method using the training model. After classification, the documents are moved to the corresponding folder in KMS using the API of Vitas/KM. The accuracy was compared with the data without feature selection and screening.