Dissertation > Excellent graduate degree dissertation topics show

Research on Product Named Entity Recognition and Normalization

Author: MeiFeng
Tutor: LinLei
School: Harbin Institute of Technology
Course: Computer Science and Technology
Keywords: Name entity corpus constructed Product named entity recognition Name entity standardization Maximum Entropy Model Conditional random field model
CLC: TP391.1
Type: Master's thesis
Year: 2011
Downloads: 20
Quote: 0
Read: Download Dissertation

Abstract


With the popularity of electronic commerce and prosperity, for the retrieval of e-commerce technology began to be more and more researchers are concerned, which, as one of the core issues of e-commerce search technology named entity recognition technology has become an important research. This paper studies the definition of the named entity corpus, named entity recognition and normalization techniques, which specifically includes the following aspects: First, the variation of the products named entity in a network environment, a product named entity an integral component of the new definition, which is conducive to a more detailed study carried out for the identification of the different components of. On this basis, the development of the product named entity corpus annotation specification, and semi-supervised methods build a quality product named entity recognition corpus. On the other hand, in order to make the product name entity standardization smooth start also gives the product named entity standardized definitions, which built a hierarchical entity library contains 21240. Second, characteristics of the structure of the product named entity divided, the product named entity recognition is divided into two stages, the first stage of identification products brand name, family name, model name and company name, identification of the second stage in the first stage based on the product named entity recognition, based on hidden Markov model, maximum entropy model, conditional random field model named entity recognition method. Feature template based on maximum entropy and conditions with airport model named entity recognition method, the product brand library and a series of library into the model used to trigger the brand name of the product, the series name and model name recognition. Experimental results show that after the brand characteristics and family characteristics into product, a 8.42% increase in F-value products named entity recognition. Finally, comparative analysis of the advantages and disadvantages between the three methods, based on conditional random field model of the product named entity recognition method to obtain the best recognition performance, the F value of 86.45%. Third, for Product Name abbreviations and more than reason caused Name entity ambiguity problem, given named entity standardized concept, and in accordance with the characteristics of the product named entity composition, given the product name of the algorithm based on the edit distance the calculation method of the degree of accuracy in the named entity standardized system reaches 84.72%. In addition, the use of extraction between adjacent product entity relation extraction method based on self-learning, and to derive the relationship between the full text of each entity based on the transport properties of the relationship, the relationship between the product entity and product names similarity calculation method standardization of the the Name entity, system accuracy of 88.09%.

Related Dissertations

  1. Research on Domain Entity Attribute and Event Extraction Technology,TP391.1
  2. Research on Extraction and Tracking of People’s Opinion,TP391.1
  3. The Frame Disambiguation of Automatic Identification of Chinese Frame,TP391.1
  4. Based on self-learning social relation extraction research,TP391.1
  5. Design and Implementation of the Chinese Webpage Classifier Based on the Maximum Entropy Model,TP393.092
  6. Based on Maximum Entropy Model Research on extracting Chinese entity relation,TP391.1
  7. Research and Implementation of Hierarchical Phrase-Based Translation Model in Statistical Machine Translation,TP391.2
  8. A Research on Chinese Word Segmentation Based on Phonetic Annotation,TP391.1
  9. The Study of Q & A System Based on Maximum Entropy Model Semantic Parsing,TP391.1
  10. E-commerce for the product summary Mining Technology,TP391.1
  11. The Automatic Identification of the Semantic Core Words for Frame Elements,TP391.1
  12. The Characteristic Analysis of the Protein Secondery Structure and Protein Interaction Prediction,Q51
  13. Automatic Semantic Role Labeling of Chinese FrameNet Based on Conditional Random Fields Model,TP391.1
  14. The Research and Application of Automatic Term Extraction Technology,TP391.1
  15. Linear chain conditional random the airport training algorithm optimization,TP181
  16. Automatic Semantic Role Labeling of Chinese FrameNet Based on Maximum Entropy Model,TP391.1
  17. Text Categorization Based on the Conditional Random Fields,TP391.1
  18. Automatically Chinese Address Recognition and Normalization,TP391.43
  19. Recognition of Chinese Prase,TP391.1
  20. Research of Intelligent Inquiry System Based on Maximum Entropy Model,TP391.6

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