WWW.DISSERTATION.XLIBX.INFO
FREE ELECTRONIC LIBRARY - Dissertations, online materials
 
<< HOME
CONTACTS



Pages:     | 1 | 2 || 4 | 5 |   ...   | 8 |

«EXTRACTION OF CONTEXTUAL KNOWLEDGE AND AMBIGUITY HANDLING FOR ONTOLOGY IN VIRTUAL ENVIRONMENT A Dissertation by HYUN SOO LEE Submitted to the Office ...»

-- [ Page 3 ] --

Most data formats of the generated structures are only segmented regions and  insufficient for representing a scene structure Recently, Saxena et al. [21] obtained a 3D scene structure from a single still 2D image using Markov Random Field (MRF) which is trained by supervised learning technique. Even though their method generates 3D models for 64.9% of 588 images, its performance depends on the trained MRF model. Z depth information is also driven from the trained MRF model. Following Saxena et al., Delarge et al. [43, 44] and Hoiem et al. [24, 45] assume that the environment is made of a flat ground with vertical walls, but this assumption generates an inaccurate structure.

–  –  –

The suggested algorithms in Sections 4 and 5 overcome the limitations listed in Table 2. The scene structure can be represented as a special graph structure (the Metaearth architecture in Section 3) and it can be generated from one scene (the methods in Section 4) and multiview images (the methods in Section 5). The suggested algorithms, presented in Section 4 and 5 do not depend on any supervised learning method, making them useful in more general application areas. Detailed theories and discussion are described in the following sections.

2.3 RESEARCH STUDIES FOR ONTOLOGY AND SEMANTICS

A model with structural knowledge has a hierarchy and semantics (Definition 3).

In this dissertation, the semantics-mapped hierarchy is called a virtual ontology (the detailed definition is described in Section 3.1). In the model with structural knowledge, semantics are represented as contextual knowledge which is created by combining virtual components in the generated structure and context.

Montague and Dowty [46] first introduced and developed semantic theory in research fields handling natural language. According to their definition, the elements of symbolic objects can be called semantic features and semantics can be a meaning.

However, many researchers define ontology according to their own concepts. For example, Anderson and Vasilakis [47] define it in terms of geometric modeling as a rigorous conceptualization of some knowledge domain. By structuring knowledge from a defined ontology, systems can semantically inter-operate when processing, exchanging or presenting information. In this concept, ontology can vary by domains and application areas.

Currently, efforts to extract ontology and semantics are a major topic of research in artificial intelligence, data mining and communication theory [48-51]. In particular, semantics and ontology play a crucial role in the development of semantic Web and Web

2.0 especially the Web Ontology Languages OWL and Resource Description Framework (RDF) [52-54].

In general, the research consists of three streams: how to generate ontology;

effective data structure; and how to intensify ontology and semantics. The first stream can be handled differently in each application. As stated above, this dissertation examines its generation from a 2D image or multi-view images.

The second stream, data format of ontology, is related to the representation of ontology. The most popular format uses a graph-based format or its own language. Goh et al. [55] use their own language called COINL via the Frame Logic method. While this kind of format is useful for a specific application, it has a disadvantage in general usage and representation simultaneously. Wu et al. [56] use a directed labeled graph to represent ontology. The graph is defined using a 4-tuple: concept node, edge for concept nodes, labels for concept node and labels for edge. The main limitation of their ontology graph is that it fails to define relationships among labels. The relationships among labels as well as among conceptual nodes are important in establishing detailed ontology. For this reason, we use Metaearth architecture as described in Section 3.

3. VIRTUAL ONTOLOGY AND METAEARTH ARCHITECTURE

This section presents the exact definitions of virtual ontology, Metaearth architecture [8] and related terms used in this dissertation. Section 3.1 explains the concept of virtual ontology and its definition in comparison with other formats. Section

3.2 describes the features of Metaearth architecture and gives a summary of Lee and Banerjee [8].

–  –  –

As discussed in section 2.3, the definition and concept of ontology varies by application and domains.

In general, ontology is defined by 5-tuple: O : {C, R, H c, rel, A0 }. O indicates “ontology”, C means “concept”, R represents “relation”, H c indicates “concept hierarchy”, rel means “function relation”, such as relationship between concepts ( R : C  C  R ) and A0 represents “axiom”, such as the relationship between C and R.

In this definition, the set {C, H C } is called “Taxonomy” and the set of Taxonomy, R, rel and A0 is called “Ontology”. However, the general representation is not suitable for virtual ontology. We apply ontology to a virtual model or a generated model after scene understanding procedures. In our concept, ontology is termed virtual ontology and

defined as:

Definition 4. Virtual ontology A virtual ontology is a structure of virtual environment with knowledge. The structure can be a hierarchy or relationships among virtual components. The knowledge of overall environment or a knowledge belongs to one or some virtual components. The knowledge is embedded in the structure with various formats.





Knowledge represented with contexts is called a contextual knowledge. In this dissertation, knowledge implies a contextual knowledge. A context is a label with one or more meanings. As shown in Figure 1, each part of a virtual robot has a label, such as basis, arm or grip, which indicates that part 1 is being used as a “Lower Basis” of the robot and part 4 is being mapped as a grip. The label defines the usage of a virtual component and guides its action. We call the label a context, and a type of knowledge in a virtual environment is represented using contexts.

Since our objective is to generate a virtual environment’s hierarchy and its contextual knowledge, the identification of all virtual components is the prior condition for establishing virtual ontology. Once the virtual components are detected, we can generate the structure, which can be a kinematic structure, a structure describing interaction, a hierarchy indicating distance, or a type of scene graph.

Figure 1 shows that the virtual robot’s structure is represented by its kinematic structure. When a mechanical part or assembled products are represented in a virtual model, it is reasonable to consider a kinematics as a structure. When using a virtual environment to describe interactions between virtual objects, the structure must indicate the degree of interaction, i.e. high and low. In most games or simulations using virtual components, the overall structure represents interactions among inner components. In this manner, the structure can be altered by the characteristics and objective of the virtual environment.

However, in a generic image such as a natural scene or photo, it is quite difficult to capture the characteristics or objective of the reconstructed virtual environment. In this case, we can use a type of scene graph as a structure. Sowizal et al. [57] define scene graph as a description of a 3D world rendered by a graphic. The format most often used is the directed acyclic graph (DAG) with nodes and arcs of various types. In general, commercial graphics software systems like VRML, OpenInventor, Performer, Java3D and X3D use this scene graph concept as a structure of overall virtual environments. In the scene graph, arcs represent relationships among nodes using parent-child relationship and reference relationship. With two types of nodes (group node and leaf node), a volume, geometry or texture can be represented. Figure 9 illustrates a scene graph in a Java3D model.

–  –  –

While a scene graph is a useful description of a virtual environment for a general

architecture, its limitations are:

When a group or patterns are repeated, the scene graph may be complex.

 The context of a virtual component can be represented as only annotation or 

–  –  –

It is not suitable for describing interactions between virtual components.

 A more effective structure for describing virtual ontology is Metaearth architecture [8]. We propose this architecture to describe any LSVE with many interactions among virtual components. It allows us to construct an LSVE with less redundant designs. The next section summarizes the characteristics and advantages.

–  –  –

“Metaearth” is a synthetic word combining meta and earth [8]. The architecture itself derives from the earth’s structure which consists of an outer core and an inner core, a mantle layer and a crust layer. The crust layer corresponds to the virtual space layer where the virtual components exist. The core layers correspond to the library layer and ontology layer. These layers are “in charge” of the assets of virtual components in the virtual space layer. The mantle layer corresponds to a mapping layer which describes the mapping from shared components in the library layer to virtual manufacturing components in the virtual space layer. Table 3 and Figure 10 describe the roles and components of the layers.

Table 3. Each layer’s role and components in Metaearth architecture [8].

–  –  –

The virtual space layer consists of multi-interactions. This graph’s vertices represent virtual components in the virtual environment. These components are replicated from the library (outer core) layer’s components. The library layer’s role is similar to the library/asset of virtual components used in the virtual space layer. Each virtual component in the virtual space layer may have interactive relationship(s) with other virtual components. This characteristic is useful for describing a LSVE. In a natural scene, this interaction relationship can be replaced by adjacent information for a scene graph. Figure 11 illustrates the interaction among some virtual components. The constructed virtual space layer is a workstation in a manufacturing facility.

–  –  –

communicating/interacting with the two conveyor belts. The relationship is represented using the interaction graph shown below the snapshot in Figure 11 (a). Next, Figure 11 (b) provides the interaction graph for a small manufacturing cell, where conveyor belts #1 and #2 are modeled using one conveyor belt model. The original model of the conveyor belt is defined in the library layer. It is replicated in two conveyor belts in the virtual space layer. This approach increases the reusability of virtual models. Figure 12 shows this mapping; the information in the mapping layer contains the rotation axes and angles, center positions’ coordinates and scale vector. This description is represented using XML format for distributed configuration between multi-virtual space layer and library layer.

–  –  –

Metaearth architecture is useful from the perspective of high reusability of virtual components and allows us to check the structure of the virtual environment. The architecture has interaction relations or hierarchical relationships in the virtual space layer and ontology information in the ontology layer. Mapping between the two layers occurs in the mapping (mantle) layer using XML format.

Note that we can expand the architecture to many large-scale environments. In this case, the virtual space layer can be divided into multi-interaction graphs. The library layers as a virtual models’ library can be divided into more detail levels and sub-layers.

Each mapping is accomplished by checking the ontology.

Metaearth architecture has eight characteristics:

–  –  –

Support for a variety of industrial areas and applications  Table 4 shows the manual Metaearth construction procedures. One objective of this approach is to generate the Metaearth architecture automatically (achieved by the suggested approaches in Section 4 and 5).

Table 4. Metaearth construction procedures and generated layer in each step [8].

In the virtual space layers, the same component and similar components are classified. Through this process, we extract the common components and use them for virtual model library/assets. Even though one virtual component is not shared with any other virtual component, we can use it as a reference model in the next LSVE. We can also expand Metaearth architecture to a significantly larger LSVE. Figure 13 illustrates the distributed virtual plants concept [8] where the distributed facilities are established effectively with the shared components via the use of Metaearth architecture.

Figure 13. Metaearth architecture for a virtual factory [8].

(a) 50 plants in virtual space (b) Interaction graph among 50 plants (c) Generation of library layer (d) Generation of mapping layer Figure 14. Generation of each layer in Metaearth architecture [8].

50 arbitrary 3D points are randomly generated as the locations of 50 plants (Figure 14 (a)). The interaction or adjacent relationship is generated as shown in Figure 14 (b). These interactions among 50 virtual components are randomly generated. We consider 8 components as shared virtual objects in the library layer. Based on this situation, we create the library layer and mapping layer as shown in Figure 14 (c) and (d). Finally, Figure 15 shows the multi-distributed virtual plant model using the Metaearth architecture (note that we have removed some of the interactions for a clearer understanding).

Figure 15. Metaearth architecture for distributed virtual plant model.

This architecture is considered as a good data structure keeping virtual ontology defined in section 3.1. As it has virtual components, their relationship, ontology / library and their mapping, various relationship and semantics can be represented in the Metaearth architecture.

Definition 5 defines virtual ontology using the Metaearth concept:



Pages:     | 1 | 2 || 4 | 5 |   ...   | 8 |


Similar works:

«Plasticity in Cu thin films: an experimental investigation of the effect of microstructure A thesis presented by Yong Xiang to The Division of Engineering and Applied Sciences in partial fulfillment of the requirements for the degree of Doctor of Philosophy in the subject of Engineering Sciences Harvard University Cambridge, Massachusetts October, 2005 © 2005 Yong Xiang All rights reserved. Thesis advisor Author Joost J. Vlassak Yong Xiang Plasticity in Cu thin films: an experimental...»

«Studies of n-type doping and surface modification of CVD diamond for use in thermionic applications MUHAMMAD ZAMIR OTHMAN UNIVERSITY OF BRISTOL SCHOOL OF CHEMISTRY BRISTOL MARCH 2014 A dissertation submitted to the University of Bristol in accordance with the requirements of the degree of Doctor of Philosophy in the Faculty of Science, School of Chemistry Word Count : 52, 223 Abstract This thesis presents the investigation of potential shallow n-type donors that are candidates to be used as...»

«ABSTRACT Title of Dissertation: ON THE DEVELOPMENT OF POSTURAL STABILITY DURING INFANCY AS A PROCESS OF GROWTH AND ACTIVE, EXPLORATORY SENSORIMOTOR TUNING Jason S. Metcalfe, Doctor of Philosophy, 2007 Dissertation directed by: Professor Jane E. Clark, Ph.D. Department of Kinesiology The process by which humans stabilize bipedal stance represents a confluence of changes associated with musculoskeletal maturation and experience-based sensorimotor learning. While investigations have documented a...»

«From Greenhouse to Icehouse: Understanding Earth’s Climate Extremes Through Models and Proxies Clay Richard Tabor A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy (Geology) in the University of Michigan Doctoral Committee: Professor Chris J. Poulsen, Chair Associate Professor Jeremy N. Bassis Associate Professor Brian K. Arbic Senior Scientist David Pollard Acknowledgements Foremost, I would like to thank my advisor Chris Poulsen for...»

«SINGLE-PARTY RULE IN A MULTIPARTY AGE: TANZANIA IN COMPARATIVE PERSPECTIVE A Dissertation Submitted to the Temple University Graduate Board in Partial Fulfillment of the Requirement of the Degree DOCTOR OF PHILOSOPHY by Richard L. Whitehead August, 2009                     © by Richard L. Whitehead All Rights Reserved   i      ABSTRACT Title: Single-Party Rule in a Multiparty Age: Tanzania in Comparative Perspective Candidate's Name: Richard L. Whitehead Degree: Doctor of...»

«THE INTERACTION BETWEEN PHYSICAL SPACE AND WAY OF LIFE IN LOW-INCOME SETTLEMENTS: CASE OF BUSTEES AND RESETTLEMENT CAMPS IN DHAKA A THESIS SUBMITTED TO THE DEPARTMENT OF INTERIOR ARCHITECTURE AND ENVIRONMENTAL DESIGN AND THE INSTITUTE OF FINE ARTS OF BİLKENT UNIVERSITY IN PARTIAL FULLFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY IN ART, DESIGN AND ARCHITECTURE By Shihabuddin Mahmud May, 2001 I certify that I have read this thesis and in my opinion it is fully adequate, in...»

«A LOCATION MODEL FOR WEB SERVICES INTERMEDIARIES By YI SUN A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA Copyright 2003 by Yi Sun I would like to dedicate this work to my family, especially my wife Ziya and our daughter Emily. Their love has supported me throughout all the hesitations and frustrations of this dissertation. ACKNOWLEDGMENTS This dissertation could...»

«AUGUSTUS LE PLONGEON: EARLY MAYA ARCHAEOLOGIST by Lawrence Gustave Desmond B.S.C., University of Santa Clara, 1957 M.S., San Jose State University, 1964 Universidad de las Americas, 1979 ~1.A., A thesis submitted to the Faculty of the Graduate School of the University of Colorado in partial fulfillment of the requirements for the degree of Doctor of Philosophy Department of Anthropology This thesis for the Doctor of Philosophy degree by Lawrence Gustave Desmond has been approved for the...»

«Throngs of Iniquity By Misop Baynun A picturesque, philosophical word path that converges with and depicts the author's life, as he sees it, through agitated, dismal, longing, hopeful, etc., etc., eyes. It's a story of pain and hope, as God perseveres in his tenacity to hold on to the hearts of those to whom he won't let go. MagPro Publishing Throngs of Iniquity Misop Baynun Table of Contents 1. Shape Without Color: Agitated longings for something 2. Blow Cool, O Merciless Winds:...»

«University of Alberta Interacting With Implicit Knowing in the Mathematics Classroom by Martina Lynn Metz A thesis submitted to the Faculty of Graduate Studies and Research in partial fulfillment of the requirements for the degree of Doctor of Philosophy Secondary Education ©Martina Lynn Metz Fall 2012 Edmonton, Alberta Permission is hereby granted to the University of Alberta Libraries to reproduce single copies of this thesis and to lend or sell such copies for private, scholarly or...»

«Essays on Expectations by Noah Swartz Smith A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy (Economics) in The University of Michigan 2012 Doctoral Committee: Professor Miles S. Kimball, Chair Professor Robert B. Barsky Associate Professor Yusuf Can Masatlioglu Associate Professor Uday Rajan © Noah Swartz Smith, 2012 For Kanua, as promised ii Acknowledgments First and foremost, I would like to thank my committee chair, Miles Kimball,...»

«PREDICTION OF TOPOGRAPHIC AND BATHYMETRIC MEASUREMENT PERFORMANCE OF AIRBORNE LOW-SNR LIDAR SYSTEMS By TRISTAN COSSIO A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA © 2009 Tristan Cossio To my friends and family ACKNOWLEDGMENTS I thank my friends and family for helping me through this whole process. I also thank my professor Clint Slatton, for his patience and...»





 
<<  HOME   |    CONTACTS
2016 www.dissertation.xlibx.info - Dissertations, online materials

Materials of this site are available for review, all rights belong to their respective owners.
If you do not agree with the fact that your material is placed on this site, please, email us, we will within 1-2 business days delete him.