원문정보
초록
영어
The high dimensionality of feature space is a big hurdle in applying many sophisticated methods to text categorization. The feature selection method is one of methods which reduce the high dimensionality of feature space. In this paper, we proposed a new feature selection algorithm based on gravitation, named GFS, which regards a feature occurring in one category as an object, and all objects corresponding to a feature occurring in various categories can constitute a gravitational field, then the gravitation of a feature with unknown category label on which all objects in the gravitational field act is used for feature selection. We have evaluated GFS on three benchmark datasets (20-Newgroups, Reuters-21578 and WebKB), using two classification algorithms, Naïve Bayes (NB) and Support Vector Machines (SVM), and compared it with four well-known feature selection algorithms (information gain, document frequency, orthogonal centroid feature selection and Poisson distribution). The experiments show that GFS performs significantly better than other feature selection algorithms in terms of micro F1, macro F1 and accuracy.
목차
1. Introduction
2. Related Work
2.1. Information Gain
2.2 Document Frequency
2.3. Orthogonal Centroid Feature Selection
2.4. Measure Using Poisson Distribution
2. Algorithm Description
2.1. Activation
2.2. Algorithm Implement
3. Experimental Setup
3.1. Validation
3.2. Datasets
3.3. Text Representation
3.4. Classifiers
3.5. Evaluations
4. Results
4.1. Results on 20-Newsgroups Corpus
4.2. Results on Reuters-21578 Corpus
4.3. Results on WebKB Corpus
5. Analysis and Discussion
5.1. Statistical Analysis
5.2. Discussions
6. Conclusion
Acknowledgment
References