Dissertation > Excellent graduate degree dissertation topics show

The Research on Construction of Conceptual Network From Dictionary

Author: HuangZuoFeng
Tutor: LuRuZhan
School: Shanghai Jiaotong University
Course: Computer Software and Theory
Keywords: Semantic relations The machine - readable dictionary Word similarity
CLC: TP391.1
Type: Master's thesis
Year: 2010
Downloads: 41
Quote: 1
Read: Download Dissertation

Abstract


Semantic information plays an extremely important role in information processing, and semantic analysis of natural language and understanding of the content, are inseparable from the support of the semantic information. Semantic Knowledge Base as manifestations of semantic information has become an integral part of basic resources in the field of natural language processing. However, most of the current semantic knowledge base is built by hand, and its size is on the order of magnitude severely accumulation and cost constraints. Quality assurance acceptable conditions, in terms of the cost of time or money cost, automated build is no doubt that the hand-built greater advantages. In this paper, how to automatically extract semantic relations from the readable dictionary. Difficult to obtain a good parser, and simply rely on character-based template matching is too rough, it is difficult to capture the complex structure of information. Therefore, this paper studies the identification method based on the specific characteristics of statistical techniques to automatically construct and to identify semantic relations. This study are as follows: First, how to construct the lexical information, syntactic information, semantic information, position information as well as some of their combination of various features type. Due to the diversity of types, spent a uniform manner for expression. In order to reduce the influence of noise, by t-test to identify the feature, and further, using a t-test to find this type of feature in the word pair. Second, characterized in order to better select the method by introducing priority priori knowledge coupled to the statistical model. And through incremental and the chance to select features and to construct the set of rules so that each rule in the rule set has a high accuracy rate, overall performance better recall rate in the set of rules. Third, because of certain factors inherent dry around, it is difficult to just from a word whether certain characteristics to determine whether there is some kind of semantic relations, so the introduction of anti-feature items. For each type of semantic relations, will construct a set of rules and anti characteristics set consisting of identification methods to identify the semantic relationships. Fourth, when using these identification methods to extract the semantic relationship instance, the semantic relationship instance to construct a conceptual network of relationships, makes a lot of words had no direct connection to an indirect connection, in order to gain greater value. Finally, in order to verify the effectiveness of the proposed method, randomly selected sample of the experimental results, the manual check. However, due to hand judgment there is a certain arbitrariness and ambiguity, it is necessary to further through a thesaurus to generate similar words and non-similar word pairs and to calculate the similarity for more objective indirect path mode Rate. This study goes a step further towards the automatically build conceptual relationships target more. If we pass the dictionary to establish a more complete and higher accuracy rate concept relationships, then we can lay a good foundation for many Chinese natural language processing applications.

Related Dissertations

  1. WordNet and the \,G254
  2. The Research on Orientation of Sentimental Word,TP391.1
  3. Research on Query Expansion & Key Technologies Based on Semantic Analysis,TP391.1
  4. A Study on the Cognate Words in "Ne" Part of Shuowenjiezi,H123
  5. New Text \,H131
  6. 1) Structure">The Study of the "Verb-object Group Followed by an Object" (V·O+O1) Structure,H146
  7. On the way of an action category,H0-05
  8. Ontology-based Geographic Information Retrieval Mechanism,P208
  9. ontology and its application in personalized information retrieval research,TP391.3
  10. The ’Yǔqí’ Construction and Some Correlative Questions,H146.3
  11. Computer-oriented Study on Syntactic and Semantic Relation between Noun1 and Noun2 in ’Noun1+Noun2’ Structure,H146
  12. A Research into the Semantic Function of "在+处所词" in Mandarin Chinese,H146
  13. Comparison of Japanese and Korean compound verb,H55
  14. Comparative Study of Chinese Culture - Loaded Words and School English Reading,H319.4
  15. Adverb \,H14
  16. The Constraints and Contrastive Analysis of Two Kinds of Double-NP Sentences,H146.3
  17. Based on the summary of the semantic relationship extraction,TP391.1
  18. \,H146
  19. A Study on Millitary Words in ZuoZhuan,H131
  20. Study on Conjunctions in Xunzi (荀子),H141

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