Dissertation > Excellent graduate degree dissertation topics show

Tree Kernel-Based Sentence-Level Sentiment Classification

Author: ZhangWei
Tutor: ZhuQiaoMing；LiPeiFeng
School: Suzhou University
Course: Applied Computer Technology
Keywords: Sentence level emotion classification Syntactic tree Convolution tree kernel Dependency tree Cutting strategy
CLC: TP391.1
Type: Master's thesis
Year: 2010
Downloads: 72
Quote: 1
Read: Download Dissertation

Abstract


With the rapid development of the Internet , the explosive growth of information on the network , the subjective texts possession ratio significantly increase . How to dig out the author's point of view is an urgent need to address the problem of analysis from these subjective texts . Emotion classification is a natural language processing technology to solve this problem , the subjectivity of its text information analysis to arrive at the viewpoint the holder emotional tendency . This paper studies the sentence level sentiment classification problem . On the detailed analysis of the basis of the importance and difficulty of the sentence sentiment classification problem , the paper proposes tree kernel function based on the sentence level sentiment classification method . The method is based convolution tree kernel function of SVM (Support Vector Machine) automatically obtain the syntactic structure information , respectively, the syntactic tree and dependency tree as a feature , and other planar features combined sentence sentiment classification . First, we explore the structural features based on syntactic tree in sentence - level sentiment classification , and proposed a SVM classifier using tree kernel and composite kernel function method to carry out the sentence level emotion classification . The experimental results show that the sentiment classification tree kernel and composite nuclear method has better performance than the linear kernel . Secondly, based on adjectives and the syntactic tree cutting strategy based on emotion word . For the former, a dynamic window algorithm to optimize a sentence containing multiple adjectives ; For the latter, add a field related to the emotional words on the classification performance . Experiments show based on the the emotional word cropped method is better than the former . In addition, experiments show that the method of classification of the hidden emotional than planar feature - based method . Finally, based on the theory of dependency the dependency tree cutting strategy , and tree kernel functions combined sentence sentiment classification method based on dependency tree . The experimental results show that the proposed dependency tree cutting strategy is effective .

Related Dissertations

  1. Research of Inquires Model Based on Weighted Grammar Dependence,TP391.1
  2. Dynamic symmetric theory : An Empirical Study Based on the mirror structure,H04
  3. Demonstrating Establishing and Management of Small Private Music Company Bandu,J609.2-F
  4. The Design and Implementation of Mining Software Repository System Based on Source Code Layer,TP311.52
  5. Automatically build the Semantic Web research in syntactic analysis,TP391.1
  6. Question answering system for text information extraction research and application,TP391.6
  7. Chinese Vietnamese machine translation experimental system,H44
  8. Tree Kernel-based Pronoun Coreference Resolution,TP391.1
  9. Research on Product Reviews Analysis Based on Appraisal Expression,TP391.1
  10. The weak stiffness structure and high-speed milling and deformation control technology research,TG54
  11. Dependency tree -based adaptive error detection and diagnosis,TP393.07
  12. Structure Analysis in Chinese-english Machine Translation,TP391.2
  13. The Research and Implemention of Image Retrieval Based on User Interested Feature,TP391.41
  14. Research of Question Answering System Based on the Analysis of Lexical and Semantic Meanings,TP391.1
  15. Research on the Classification Based on the Reconstruction of Solder Joint,TP391.41
  16. Research on Structure Transition Technology for SMT,TP391.2
  17. The Fatigue State Recognition of the Driver Based on Eye Detection,TP391.41
  18. Research on Image Compression and Implementation Using TMS320C6713 Based on SPIHT Algorithm,TP391.41
  19. Research on Joint Target Detection for Dual-Sensor Image and System Implementation,TP391.41
  20. The Study of Text Index Construction for Large-Scale Dynamic Collection,TP391.3
  21. Video Coding Research Based on Texture Characteristics,TP391.41

CLC: > Industrial Technology > Automation technology,computer technology > Computing technology,computer technology > Computer applications > Information processing (information processing) > Text Processing
© 2012 www.DissertationTopic.Net  Mobile