Dissertation > Excellent graduate degree dissertation topics show
A Study on the Method of Feature Selection in Text Categorization
Author: SongLiPing
Tutor: ZhangXiaoYan
School: Xi'an University of Science and Technology
Course: Applied Computer Technology
Keywords: Text Categorization Semantic concept Feature Selection Weight calculation Vector space model
CLC: TP391.1
Type: Master's thesis
Year: 2009
Downloads: 28
Quote: 2
Read: Download Dissertation
Abstract
|
Text classification is the basic functions of the large-scale text processing , but also improve other text-processing functions and an effective means of quality . Text classification, by category text storage , retrieval and further processing . Therefore, the text can be classified quickly and effectively has become an important research topic . In text classification , feature space dimension up to tens of thousands, too large feature space will result in the assessment of the characteristics of the sample statistic becomes difficult , thus reducing the generalization ability of the classifier from the original feature set selected the most representative characterized in that the very necessary . Effective feature selection can improve the efficiency and classification performance of the classification task . In this paper, the core technology analysis of a typical text classification system , the system structure based on the analysis method based on semantic concept . The concept of semantic analysis method can be seen as an extension of the vector space model , feature extraction algorithm by combining the concept of Hownet word , the word space is mapped to the concept of space , through the merger of polysemy disambiguation and synonym , to drop dimension purpose, and as far as possible to achieve orthogonality between words , the text keyword with a smaller semantic space means , so that the text in the new generation of semantic space closer . In addition, on the basis of the traditional TF - IDF the right weight calculation algorithm for the consolidated consider the position of the characteristic words , the support degree of semantic factors and characterized in the same occurrence frequency , category strength of the semantic concept , enhancement of the characteristic of the text content of the performance and distinguish ability , and a combination of TF-IDF semantic factors and semantic concept weight improved algorithm is applied to the classification system . In this paper, the design and realization of a Chinese text classification system , experiments improved features choose and the right to re- calculation algorithm with traditional statistical algorithm carried the comparative analysis , experimental results show that the improved characteristics selection algorithm and the right to re- calculation algorithm of classification performance has some of the improved, with higher precision and recall rate .
|
Related Dissertations
- The Discovery of User Concept Region Based on Multiple Instance Learning,TP391.41
- Research on Text Classification Based on Biomimetic Pattern Recongnition,TP391.1
- Feature Extraction, Selection and Combination in Lipreading,TP391.41
- Research on Focused Crawler Based on SVM Classification Algorithm,TP391.3
- Research on Feature Selection and Construction in Emotion Speech Recognition,TP18
- Based on Data Distribution Characteristics of Text Classification,TP391.1
- Research and Implementation of a Dynamic Feature Selection Method for Vehicle Recognition System,TP391.41
- Research on Face Recognition Based on AdaBoost Algorithm,TP391.41
- Research on Feature Extraction, Selection and Classification Algorithms for Pulmonary CAD,TP391.41
- Design and Implementation of the Character Classification System Used in Search Engine,TP391.3
- Research and Implement of the Theme Crawler for Automotive Industry,TP391.3
- Research and Application on Short Message Text Clustering,TP391.1
- Research for Event Extraction Method in Specific Domain Based on Tree Conditional Random Field,TP391.1
- Online Education News Text Categorization System Design and Implementation,TP391.1
- Forum message text for effective data extraction studies,TP391.1
- SMBSDD: An improved search mechanism in unstructured P2P networks,TP393.02
- Web-based social network search method to judge the identity of names,TP393.09
- Web news hot discovery system design and implementation,TP393.09
- Based on swarm intelligence optimization algorithm for medical image feature,TP391.41
- FSVM -based data mining method and its application to intrusion detection research,TP393.08
- Research on multi-role social networks,TP393.09
CLC: > Industrial Technology > Automation technology,computer technology > Computing technology,computer technology > Computer applications > Information processing (information processing) > Text Processing
© 2012 www.DissertationTopic.Net Mobile
|