Dissertation > Excellent graduate degree dissertation topics show

Researeh on Opinion Extraetion of Chinese Produet Review

Author: LiChunLiang
Tutor: ZhuYanHui
School: Hunan University
Course: Applied Computer Technology
Keywords: Product Review Opinion mining Emotional words Attribute characteristics Polarity
CLC: TP391.1
Type: Master's thesis
Year: 2011
Downloads: 20
Quote: 0
Read: Download Dissertation


With the development of e-commerce and web 2.0 applications , more and more consumers after the purchase and use of products , like products viewpoints attitude in the e-commerce website , forum , blog , these comments contain user characteristics , views of features, performance , consumer before purchase always consulting others views on the product in order to make informed purchase decisions , manufacturers can also be based on user reviews to improve the product manual to browse these massive product reviews time-consuming and inefficient , and there are lag and one-sidedness . Recently, unstructured network product reviews automatically opinion mining has become a hot research topic . In this paper, the emotional perspective extract resources , product attributes feature extraction , the attribute feature words with emotional word identification and polarity determination to carry out the in-depth study and research work of this paper are as follows : (1) the use of open source the the tools Larbin and Xpath mobile channel for shopping site directed reptiles , and page format using Xpath metadata extraction , and ultimately build a cell phone reviews corpus . ( 2) in building emotional point of view to extracting resources , the proposed construction method based on the Baidu encyclopedia emotional Dictionary conjunctions dictionary - based and dependency syntactic relations combined field of emotion dictionary method network emotional Dictionary , the emotions modified dictionary construction method . ( 3 ) characteristics of the product attributes extracted aspects , the proposed rule-based and statistical recognition algorithm and recognition algorithm to extract the product attribute characteristics CRF - based attribute characteristics to improve the the former accurately rate of 0.56 , the coverage rate of 0.73 , while the latter is the accurate a higher rate, was 0.78 , but the coverage is only 0.46, in order to be compared with other researchers , Hu and Liu's method is applied to the experimental environment , experiments show that the two methods of this article in the method of Hu and Liu . (4) with the identification and polarity attribute characteristics and emotional words judging the accuracy with SVM recognition algorithm , and matching algorithm with the nearest neighbor recognition algorithm based on dependency syntax with comparative experiments , SVM with the recognition algorithm reached 0.83 , the coverage rate of 0.62 F 0.71 , much higher than the other two methods , to obtain the best performance .

Related Dissertations

  1. Research on the Topologies of Polarity-Variable Micro Oxidation Pulse Power,TN86
  2. Design and Realization of Text Removal and Repair System in Digital Image,TP391.41
  3. Research on Sentiment Orientation Analysis of Blog Article Based on Blog Search,TP391.1
  4. Research on DBD-Based the Space Periodic Boundary,TM83
  5. Converter transformer winding end analysis and optimization of the electric field,TM42
  6. Based naphthalene fluorescent molecule synthesis, characterization and photophysical properties,O621.3
  7. Opinion Mining in Online Forums for Financial Q & A System,TP391.3
  8. Study on Sentiment Polarity Analysis of Subjective Text,TP391.1
  9. A Novel Microsyringe Extractor for Determination of Various Polar Chemicals from Plant Sample,O658.2
  10. Research on Internet Hotspot Analysis and Monitoring Technologies Based on Topic Detection and Sentiment Analysis,TP393.09
  11. Research on Computation of Chinese Text Sentiment Classification of Multi-granularity,TP391.1
  12. Prelimilary Experimental Study of Three-demensional Engineered Skeletal Muscle Tissue in Vitro,R318
  13. Study on the Experience Value of Creative Household Products,J525
  14. The Teモ\,H364
  15. Processing of Affective Word during Sentence Reading: An ERP Study,B842.1
  16. Novel Fluorescent Viscosity and ROS Probes Based on Tuning ICT Effect of D-pai-A Fluorophores,O641.1
  17. Design of Illegal Information Filtering Firewall with the Emotional Words,TP393.08
  18. Research of High-Power Novel PWM Induction Heater Based on DSP,TM924
  19. The Inner Electrical Field Analysis for Converter Transformer and the Testing System Design for Oil-paper Insulation,TM422
  20. Different emotional word item memory and source memory ERP Study,B842.3

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