Supervised Learning Technique using TF-IDF for Text Classification

カテゴリ: 部門大会

論文No: GS13-7

グループ名: 【C】平成21年電気学会電子・情報・システム部門大会講演論文集

発行日: 2009/09/03

タイトル(英語): Supervised Learning Technique using TF-IDF for Text Classification

著者名: Mohammad Golam/ Sohrab Mohammad Golam/ Sohrab (徳島大学),Mohamed Abdel/ Fattah Mohamed Abdel/ Fattah (徳島大学),Motoyuki Suzuki(徳島大学),Ren Fuji(徳島大学)

著者名(英語): Sohrab Mohammad Golam (The University of Tokushima),Fattah Mohamed Abdel (The University of Tokushima),Suzuki Motoyuki(The University of Tokushima),Ren Fuji(The University of Tokushima)

キーワード: Text classification|Feature Selection|Data mining|Information retrieval|Machine LearningStemming

要約(日本語): Text classification is defined as the task of classifying documents into a fixed number of predefined categories. To improve the performance of text classification, it is desirable to reduce high dimensionality of the feature space. Thus the original documents can be represented as a series of vectors. In this paper, we use TF-IDF (Term frequency-Inverse document frequency) as a feature selection criterion, in order to measuring the importance of a term in a document and to improve the recall and the precision of the retrieved text. Experiments are performed over self collected data corpus with six categories of 200 documents. The proposed approach is a supervised learning technique and deals with text string representation of a document and is transformed into a numeric feature vector at a pre-processing step and then the vector is provided to a learning algorithm as input.

PDFファイルサイズ: 2,315 Kバイト