Characristic Analysis Of Association Semantic Link And Its Application On The Web

Characristic analysis of Web link is the key problem of theory and technology ofthe network application. At present, Web hyperlink analyses have been studied, e.g.classic algorithms such as Pagerank, HITS. With the massive growth of webinformation, Semantic Web (SW) is becoming the research hotspot in webinformation processing. Semantic link analysis needs to be deeply studied byresearchers. In this thesis, we analyze the link characristic of Association LinkNetwork (ALN), which is a kind of Semantic Web focusing on association-semantic.Based on the theories and methods of Web link analysis, graph theory, this thesisanalyzes characteristics of Web resources-level links and keywords-level links,explores the generating mechanism of ALN topology and its semantic application.This thesis aims to provide theoretical support and practical reference for the Webintelligent activties such as intelligent browsing, association learning andrecommendation of domain knowledge from Web resources.The main contents of this thesis are organized as follows:1. To master the complexity of association semantic links among Web resources,we do semantic link analysis in Web resources-level. According to the weights ofsemantic link, some associated semantic links are filtered to find the changingtrends of the statistical characteristics of association semantic link.1) The changing trends of average path length of ALN are found by analyzingthe path characteristic analysis of association semantic links.2) The changing trends of semantic clustering characteristic of ALN are foundby analyzing the clustering coefficient of association semantic links.3) The degree distribution properties are discovered by analyzing the degreedistribution characteristics of association semantic links.2. The rapid mining model of sparse association semantic links is presented to solvethe problem that the sparse association semantic links are easily discarded. It canefficiently reinforce the completeness of association semantic of domain Webresources.1) From the dimension of time effectiveness, we analyze the characteristics ofassociation semantic link in keywords-level to find why the sparseassociation semantic link is easily discarded.2) Further, we present the interval existence theory of the sparse associationsemantic links, and propose a rapid mining model of the sparse associationsemantic links. In addition, we evaluate the validity of this model.3. The layer-division model based on semantic concentricity is proposed to solvethe hot topic discovery in domain Web resources by analyzing characteristics ofassociation semantic link in keywords-level.1) The document frequency distributions of four types of keywords are discussed based on the active traction and passive traction of keywords todiscover the function between keywords number and document frequency.2) Further, we propose the layer-division model based on semanticconcentricity to provide the fundamental theory for the topic searching onthe association semantic environment. In addition, we evaluate the validityof this model.3) Based on the proposed layer-division model, the current attention topic andlong-term attention topic within a domain are discovered to assist/guidemastering the domain knowledge for users.4. Based on the above statistical characristics analysis in Web resources-level andthe semantic characristics analysis in keywords-level, we study the generatingmechanism of ALN topology, to establish the ‘bridge’ from the semanticcharacristics to the statistical characristics.1) The link out-degree calculation model based on active traction of keywordsis presented to establish the mapping relations between content of Webresources and its link out-degree.2) The link in-degree calculation model based on passive traction of keywordsis presented to establish the corresponding relations between content of Webresources and its in-degree.5. Two applications based on characristics analysis of association semantic links arepresented.1) According to the simple idea of “new source tracing”, we propose anintelligent browsing model of domain Web resources based on associationknowledge flow to provide the browsing path of several associatedproblems for users.2) According to the simple idea of “learning from Web resources”, we proposethe Web resources-oriented association learning model to provide thefreshness, abundant knowledge of a given topic for learners.This study can provide solutions and research methods based on associationsemantic for Web intelligent acitives such as the intelligent browsing of Webresources, the association learning of domain knowledge from Web resources. Thisstudy also can provide theoretical foundation for Web resources recommendation,text semantic search, E-learning based on association-semantic. It can relieve thecognitive burden of users in a certain extend when they aquire domain knowledgefrom the Web.