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

  1. The Discovery of User Concept Region Based on Multiple Instance Learning,TP391.41
  2. Research on Text Classification Based on Biomimetic Pattern Recongnition,TP391.1
  3. Feature Extraction, Selection and Combination in Lipreading,TP391.41
  4. Research on Focused Crawler Based on SVM Classification Algorithm,TP391.3
  5. Research on Feature Selection and Construction in Emotion Speech Recognition,TP18
  6. Based on Data Distribution Characteristics of Text Classification,TP391.1
  7. Research and Implementation of a Dynamic Feature Selection Method for Vehicle Recognition System,TP391.41
  8. Research on Face Recognition Based on AdaBoost Algorithm,TP391.41
  9. Research on Feature Extraction, Selection and Classification Algorithms for Pulmonary CAD,TP391.41
  10. Design and Implementation of the Character Classification System Used in Search Engine,TP391.3
  11. Research and Implement of the Theme Crawler for Automotive Industry,TP391.3
  12. Research and Application on Short Message Text Clustering,TP391.1
  13. Research for Event Extraction Method in Specific Domain Based on Tree Conditional Random Field,TP391.1
  14. Online Education News Text Categorization System Design and Implementation,TP391.1
  15. Forum message text for effective data extraction studies,TP391.1
  16. SMBSDD: An improved search mechanism in unstructured P2P networks,TP393.02
  17. Web-based social network search method to judge the identity of names,TP393.09
  18. Web news hot discovery system design and implementation,TP393.09
  19. Based on swarm intelligence optimization algorithm for medical image feature,TP391.41
  20. FSVM -based data mining method and its application to intrusion detection research,TP393.08
  21. 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