Font Size: a A A

The Model Of Text Concept Semantic Space And Its Applications

Posted on:2015-10-26Degree:DoctorType:Dissertation
Country:ChinaCandidate:X WeiFull Text:PDF
GTID:1228330434459461Subject:Computer application technology
Abstract/Summary:PDF Full Text Request
Compared with keyword, concept has bigger semantic granularity and holdsmore semantic information, which is used in ontology construction, text semanticrepresentation, semantic annotation, semantic search, etc., in order to improve theefficiency of text semantic processing. Therefore, the effectiveness of conceptsemantic processing will directly influence the effectiveness and efficiency of textsemantic processing. In the area of text-faced concept semantic processing, most ofresearches focus on the specific algorithms of concept semantic processing, whilefew of them aim at the basic problems of concept semantic, such as, What s thecondition of textual concept forming? How many keywords are economical torepresent a concept? What are the powers and rules of concept semantic evolution?All these questions are related to the basic rules of concept semantic movement,which provide methodology guidance for specific algorithms of concept semanticprocessing. So research on the basic rules of concept has the same importance asthat on specific algorithms.Facing the specific application area of Web text semantic processing, this paperuses the essential idea of Thermodynamics Laws and Dissipative Structure forreference to construct Text Concept Semantic Space (TCSS) as the space of conceptmovement and evolution. By researching the TCSS from both macro and microperspectives, this paper preliminarily discovers some basic rules of concept semanticmovement and evolution, which can be used as the support and reference for allkinds of textual concept semantic processing. The main research contents of thispaper are listed as follows:1. This paper discovers the similarities between thermodynamics system andconcept semantic system by making the analogy between them, accordinglyproposes the Text Concept Semantic Space (TCSS), analyzes basic features andDissipative Structure of TCSS according to thermodynamics system, and makesthe connections between thermodynamics system and TCSS. All the aboveresearches establish the theory basis for the research on TCSS.2. Based on the quantitative analysis of TCSS, this paper proposes three basicproperty theorems(Calculating the information content of TCSS, Conservationof Information content of TCSS, Information content convergence of TCSS),which preliminarily reveals the basic rules of text concept semantic movement.Based on the above basic theorems, this paper puts forward a method ofconstructing TCSS without priori knowledge and manual intervention. Thisconstruction method can also be used in concept extraction, ontology automaticconstruction, text semantic annotation, etc. Some experiments in TCSS construction are also the verification to the proposed basic theorems.3. This paper takes advantage of little priori knowledge from the domain tooptimize the construction processing of TCSS in order to increase the precisionof TCSS, which can provide better support for TCSS-based applications. Thispaper first proposes a common algorithm optimization model based on prioriknowledge, which is used to provide methodology guidance for the optimizationof each step of the TCSS construction, then uses the model to optimize mainsteps of TCSS construction with the guidance of priori knowledge, such asincreasing the accuracy of keywords extraction algorithm, increasing theaccuracy of association rules mining algorithm, optimizing the construction ofkeywords association semantic link network, increasing the accuracy of concepttree construction algorithm. All the optimizations on the above steps realize thewhole optimization of TCSS.4. Based on Dissipative Structure Theory, this paper builds the DissipativeStructure model of TCSS in order to research the process of the formation andsemantic evolution of a concept. This paper proposes qualitative and quantitativemethods to judge if a TCSS forms the Dissipative Structure, preliminarilyreveals the impetus of concept evolution in TCSS, analyzes the macro and microevolution phenomenon in the process of concept evolution, and preliminarilyreveals the basic rules of concept evolution. Based on the above researches, thispaper discusses how to detect unconventional emergency based on TCSSDissipative Structure.5. This paper discusses two applications of TCSS.(1) Construct a large-scaletesting dataset for text semantic annotation. To solve the problem that the scaleof test dataset of research on text semantic annotation is relatively small andcan t provide comprehensive evaluation on annotation algorithms, based on theidea of TCSS, this paper uses Mesh and PubMed as data source and constructs alarge-scale testing dataset for text semantic annotation. Based on the dataset, thispaper also provides some evaluation references, which can support theresearches on semantic annotation.(2) Extract Facet automatically based onwebpages set. Based on the idea of TCSS, this paper proposes aMultidimensional Semantic Index (MDSI) of webpages, which can create asemantic-rich index for webpages. Facets are extracted by analyzing semanticmapping relations in MDSI, which solves the problem of automatic facetextraction on massive unstructured texts to some extent and can support the facetsearch.Research results in this paper preliminarily reveal the fundamental rules ofconcept semantic movement, and provide some specific methods and algorithms forthose researches on concept, which can be directly used in text-faced conceptextraction, concept representation, ontology construction, text semanticrepresentation, text semantic annotation, semantic search, etc., in order to improve the effectiveness of Web applications.
Keywords/Search Tags:Textual Concept Semantic Space, Thermodynamics System, Dissipative Structure, Concept Semantic Evolution, Large-scale Dataset, FacetedSearch
PDF Full Text Request
Related items