Developing a Visual Lexical Model for Semantic Management of Architectural Visual Data Design of Spatial Ontology for Caravanserais of Silk Roads Elham Andaroodi(*), Frederic Andres(**),Kinji Ono(***), Pierre Lebigre(****) National Institute of Informatics (Japan), Ecole dArchitecture Paris Val de Seine (*) Ph.D. student, National Institute of Informatics (**) Associate Professor, National Institute of Informatics (***) Professor Emeritus, National Institute of Informatics National Institute of Informatics, Hitotsubashi 2-1-2-1407 Chiyoda-ku, Tokyo,101-8430, Japan (****) Professor, Ecole dArchitecture Paris Val de Seine, 14, rue Bonaparte 75006 Paris
[email protected] ,
[email protected] ,
[email protected] ,
[email protected] )>IJH=?J In this article we discuss issues related to developing a visual model in order to conceptualize and represent lexical knowledge of the historical architecture domain to improve the semantic management of visual information and reduce the ambiguity problem of data access. This model -domain ontology- provides systematic lexical specification of the components and the relationships between them that reflects 3 dimensional characteristics of the architectural content for a typology of historical buildings, caravanserais of the Silk Roads. This paper discusses issues about design and data input of the lexical model, specially the multilingual terminology. Further as the process of defining relationships is the main focus of developing a spatial ontology model, this paper will review the necessary background knowledge and the contribution of architectural domain for collecting and creating relationships in order to cover the attributes of architectural space and the constraints and properties that will influence defining pair of links for the corpus of this research. The initial design of prototype of this model with relationships using protégé tool is also presented. Finally we conclude with a technical application of this knowledge model which purposes are to support the semantic of visual architectural information and to enhance data management for the users for internet based access.
of the content. This model needs also to provide advanced information access for the users without any ambiguity regarding the content (e.g. different perspectives and shapes of the components, technical knowledge of details, etc).
Key words: Visual lexical model, spatial ontology, semantic management
2.1. Background
Reviewed and accepted: 12 August 2004
The first step for capturing the semantics of 3D architectural data is to develop a knowledge representation model of it capable enough to cover lexical knowledge and to reflect its spatial characteristics. Ontology seems to be a proper concept in NTIC 2 to visualize architectural knowledge. It is originated from philosophy and is the study of the kinds of things that exist. But the definition given by Gruber to ontology, as a specification of a conceptualization [1] focuses on its new role in knowledge management. It represents a declarative model of the terms and relationships in a domain3 and provides lexical knowledge-base. As a first step domain ontology tries to present a taxonomic terminology of the content. Further it provides the relationship links between components based on their conceptual characteristics [2]. Such lexical model of domain ontology can conceptualize the visual databases like architectural data.
1.
Introduction
Management of cultural heritage information systems mostly confront with multi disciplinary content of cultural databases. Factors such as history, time, location, space, social behavior, tradition, etc. will provide a wide variety of data and make the process of data gathering and distribution complicated. Therefore developing proper models for each special cultural field in order to conceptualize the content, to specify and represent its technical knowledge seems essential. Such models need to share common understanding of the structure of information and introduce standards to allow data interoperability between different systems and communities, especially over internet. As a subset of cultural heritage, creating a knowledge model for architectural remains will face the main characteristics of a building, three-dimensional form and spatial organization of components as a whole. This characteristic is mostly reflected in visual data (3 dimensions information such as photos or 3D models) that record a historical building from different angles or views and show its components from different perspectives. Such data based on their specific content contain many technical details of the domain. Fig1 and Fig2 show 2 examples of such databases, different perspectives of main façade of a subset of historical buildings, a caravanserai. For management of such data and mostly annotation of images, technical domain knowledge within a spatial model that is proper for describing an architectural content is needed. We can see that components have different shapes in these 2 photos. There fore such a systematic model needs also to be perspective independent. Based on the above example, the main challenges ahead for data management of such architectural visual content is to search and create a systematic knowledge model that covers the spatial features
Journal of Digital Information Management
This paper presents results from an on-going research1 focusing on the development of a semantic knowledge model, architectural domain ontology. This model conceptualizes the domain knowledge and presents a lexical visual specification of the architectural content to reflect its spatial characteristics, which means the hierarchical structured term-sets to name components and the relationships between them. The case study of historical relics in this research for designing ontology is caravanserais of Silk Roads. One application of this model (which is subject of a related research) tries to facilitate accesses to architectural visual data such as images or 3D models and to help the end user for enhanced data acquisition over internet. In the following parts of the paper, we introduce the ontology model developed for caravanserais. We discuss issues related to design process of lexical multilingual knowledge base in section 2. The spatial ontology and visual model through defining relationships is discussed in section 3. Finally, section 4 concluded with some discussions on application of ontology model for enhanced visual data management. 2. Developing a Lexical Model for a Subset of Cultural Heritage
2.2. Designing Ontology as a Case Study on Architectural Data Based on the characteristics of ontology and the model it provides, this research is focusing on developing a domain ontology for a special typology of architectural heritage, caravanserais of Silk Roads. 1
2 3
The research is conducted in National Institute of Informatics (NII) in Japan in cooperation with the architecture school of Paris Val de Seine (EAPVS) in France under the “Digital Silk Roads Initiative Framework” (DSRIF) in cooperation with UNESCO, as part of a PhD. study of the SOKENDAI university which is supported by the scholarship. New Technologies of Information and Communication Poli-Roberto, Framing ontology-second part http://www.formalontology.it/Framing_second.htm
q Volume 2 Number 4q December 2004
151
Figure 1: Courtyard facade of Deir-e-Gachin caravanserais from the entrance, view from front
Figure 2: Lateral courtyard façade of Deir-e-Gachin caravanserais
Figure 3: A snapshot of the ontology developed in protégé 2000. The classes and Persian term-set subclass and the properties as slots.
Journal of Digital Information Management
q Volume 2 Number 4q December 2004
152
This case study focuses on caravanserais, historical buildings constructed along Silk Roads. Their main function is to support caravans life [3]. They were built in a chain in such a way that as Robert Hillenbrand said they punctuated the major historic overland routes at intervals of a days journey [4]. One of the goals for developing an ontology knowledge model for caravanserais of Silk Roads is to enable semantic management to related architectural information and enhanced digital data documentation. 2.2.1. General Process of Designing Ontology The process of designing ontology depends on the type of it and the domain knowledge [5]. But in practical terms, N. Noy and D.L. Mc Guinness [6] provided the following steps regarding the development of ontology:
Defining classes in the ontology,
Arranging the classes in a taxonomic (subclasssuperclass) hierarchy,
Defining slots and describing allowed values for these slots,
Filling in the values for slots for instances.
The creation of a knowledge base is done by defining individual instances of these classes and by filling in specific slot, value information and related restrictions. The first key issues in designing a domain ontology considering the above mentioned general process are to define the content of domain knowledge and in the way to collect this content. This paper provides some solutions based on the case study of the architectural domain ontology.
etc. (Fig 3, 4). In a first step, this terminology set has populated with terms from Central Asian languages (e.g. Persian, Arabic) and then it has been extended to 9 languages in order to cover the needs of multi-lingual users9 (Fig 5). Metadata standards or thesauruses (Art & Architecture thesaurus [9] for English term-set and the database thesaurus of ministry for the Culture and Communication - Direction of Architecture and Inheritance for French language [12]) have been added to the taxonomy of term-set. It helps to disambiguate the meaning of terms. They give the hierarchical structure to terms set of the ontology as it is shown in Figure 6. Each hierarchy in the thesaurus is defined in the ontology as a class and each related term is attached to it as an instance of term-set. In this figure, you can see that the term brick as an instance of the class English term-set is linked to its hierarchical classes such as: clay products, clay and clay products, inorganic material, materials by composition, materials, material facet (from bottom to top of the hierarchy). The process of increasing the technical terminology, extending it to other languages and giving hierarchical structure to term-set in this domain has several challenges ahead in which some are listed as below[13]. -
As there is no unified reference of dictionaries for collecting technical terms in the domain of caravanserais, references such as glossaries related to historical architecture of Central Asia in various languages including Persian have been studied. This study enabled to extract proper terms.
-
Finding proper equivalents for technical architectural terms mostly face multi-lingual ambiguity according to their usages. In some cases, even no equivalent in other language can be found for a special term that names a component in a caravansary of a special region. For example a component in case study of caravanserais -as a term related to a space with the shape of 8 angels just after the entrance that acts as a joint for distribution of passengers and a filter between inner space and outer space- is called hashi in Persian language but there is no single word as equivalent in other languages.
2.2.2. Issues on the Process of Designing Lexical Model as Part of Ontology for Caravanserais of Silk Roads Following the above questions, the process of developing domain ontology is started with capturing the domain knowledge or the related terminology. Generally terminologies of domain ontology are taken from resources such as dictionaries or lexical references (like wordnet [7]), metadata standards (like VRA [8]), Thesauruses (such as AAT [9]) or upper ontologies (like SUMO 4[10]). The focused domain (i.e. caravanserais of Silk Roads) contains specific terminology that goes over the above references. As this typology of historical architecture focuses on Central Asia, technical terms are collected after a complete field survey and recognition of components of the type of buildings and theory study of the related references and their glossaries5. The collected technical terms that name components of the corpus of this research 6 has been completed in a knowledge base terminology stored in RDF7 format. We use the knowledge engineering tool, Protégé 2000. This tool supports the construction on ontology in a frame-like fashion including classes and slots [11]. Classes define hierarchies, and slots contain values or information for each instance of a special class. This tool covers the needs of multilingual lexical databases by using Unicode coding standard8. The design process of caravanserais ontology starts with multilingual terminology including lexical information of each term -as an entity or componentas subclass of caravanserais terminology. Each term is an instance of the related class. We added properties to each class in slots with different values in order to annotate terminology (instance of classes) with lexical properties such as definition, pronunciation, 4 5
6 7 8 9
10
11
Some standards for multilingual equivalents or related references such as the General Ontology for Linguistic Description [14] or ISO standard 5964 [15] can help this process10. One issue is the lack of multilingual equivalence standard on the field of historical architecture11. -
An ID structure is studied to connect the different multilingual equivalents in such a way that no language is used as bridge. Also the languages have the same level of priority.
-
Thesauruses are language dependant. Therefore each language term set is connected to its related thesaurus following an interlingua representation approach. For example as it is shown in Figure 6, English term-set is bridged over AAT [9]. Meanwhile thesaurus does not exist for some languages.
2.2.3. Knowledge Model of Spatial Ontology The ontology model of this research can be defined as follows: Ontology model = {Multi Lingual Term-Set + Multi Lingual Thesaurus + ID set of equivalences + Relationships}
Suggested Upper Merged Ontology http://ontology.teknowledge.com/ This domain knowledge is acquired within a research done by the first author on Iranian caravanserais, case study of desert out city samples along Silk Roads. The related projects to support the content are “Inventory of Caravanserais in Central Asia” project of UNESCO under general Coordination of Prof. Pierre Lebigre, Professor of Ecole d’Architecture Paris Val de Seine. Desert Out-city Caravanserais of Silk Roads in Iran mainly belonging to 16th till 18th century. Resource Description Framework http://www.w3.org/RDF/ http://www.unicode.org/ The cooperation of the experts of UNESCO has been appreciated here for extending the term-set to Russian, Azeri, Italian and Chinese and for checking Arabic and French terms. For further information about metadata multilingual thesauri please refer to: http://jodi.ecs.soton.ac.uk/Articles/v01/i08/Hunter/ The main target of this research for multilingual equivalence is to reach to a consensus for naming a component. Issues related to different types of equivalence (like exact, inexact, partial, single to multiple, etc.) specially in historical architecture multilingual terminology needs more research outside the scope of this study.
Journal of Digital Information Management
q Volume 2 Number 4q December 2004
153
Figure 4: A snapshot of the English term set class and the related data. The different languages are as classes, each term as an instance of the class, and the explanation about each instance as a slot with different value type.
Figure 5: Architecture of Multi-lingual Terminology Ontology Support
Journal of Digital Information Management
q Volume 2 Number 4q December 2004
154
Figure 6: A snapshot of the English term set class and the hierarchical structure of the Art and Architecture Thesaurus.
Figure 7 : Snapshot of Slots of English term-set class and related value and cardinality of slots
Figure 8: The graph of classes and sub classes of terminology and multilingual equivalents using IDs which is created by a plug in of
Protégé, TGViz for visualization of lexical data
Journal of Digital Information Management
q Volume 2 Number 4q December 2004
155
The first 3 parts of this model conceptualize the architectural content through recognizing its components and the terms in a taxonomic way, by assigning a key to each component. Further, it uses a single ID to connect equivalent IDs of different languages (Fig 8). These 3 parts only provide the lexical knowledge base for visualization of the components as it is needed in the domain of historical architecture, by defining relationships between them. We discuss, in the next section, about the spatial characteristics of the architectural content as part of the ontology model. 3.
Visualization of Lexical Model and Design of Spatial Ontology
But as an architectural space is formed by the integrated components which mainly are reflected in 3 dimensional spaces, This lexical knowledge-base is completed by a spatial annotation through defining relationships between these components in a visual model, which is discussed as follows: 3.1. Defining Spatial Relationships in Ontology Model 3.1. 1 Background Knowledge As it is mentioned in architectural domain ontology the main characteristics of architecture, which means spatial organization, structural properties and environmental values are mainly reflected in 3 dimensional spaces. This process tries to cover these characteristics by defining relationships based on the way the components or elements are interrelated, interdependent or interacting with each other like a system 12 in the building, and according to structural, functional, temporal or spatial composition of components in the case study of caravanserais. The main challenge of this process is how visual information which is perceived directly in a space in caravanserais can be translated into verbal Natural Language information by ontology model. That means how we can reach to a systematic lexical description of a space using ontology structure in such a way that the values of a space, like form, geometry, circulation, material, structure, color, light, etc. can be described and can be linked to visual data (like images) by ontology model. This process is generally subject of the domain of cognitive science or psycholinguistics, such as how language structure space [16] and language and spatial cognition [17] which tries to find answer to this general question: how we can talk about what we see in space? For making a representation of space as A.G. Cohn addresses [18], Questions have to be addressed regarding the kind of spatial entity being used (e.g. regions, points), and the way of describing relationships between these entities (e.g. their topology, size, distance, orientation or shape). The main practical usage of this research for the ontological representation of space is the design of relationship links between components or terms in the caravanserais in order to reflect the attributes of space. Spatial relation is the subject of researches on systems of qualitative representation of space which mainly try to define spatial relations for qualitatively distinguish orientation and description of entities of space. The application of spatial reasoning is in cognitive science, geographical reasoning, data-base oriented research concerned with encoding pictorial information and Artificial Intelligence for bidirectional communication with Robots [19]. Although application of this research, a spatial lexical model for historical
13 14
3.1.2. Categories of Relationships Definition: A relation is a function to link 2 terms as components or entities. The general category of relationship is defined as R. R is Union of spatial, upper-level, construction and environment relationships: R:{S U U U C U E} è Spatial Relationships
The ontology which is designed in this research goes further than a multi-lingual lexical database and tries to provide a visual model of terms in the domain that helps with a spatial graph of the terminology. Based on what is mentioned in section 2.2 as the preliminary stage each component in an architectural relic is presented in the ontology tool (protégé) as a word, an instance of the related language class. It has independently some properties that we input the information by defining slots (Fig 6, 7).
12
architecture is not found directly in the available references, but the defined relationships are close to those needed for this ontology. These relationships are defined in the below mentioned categories.
Definition: A spatial-type relationship, S, between 2 terms as entities is defined by the following relations: S : { Distance: far from, near to, close to, next toOrientation: above of, upper of, top of, below of , lower of, bottom of, interior of, posterior of, exterior of , behind of, in front of, left of, right of, north of, south of, east of, west ofTopology: inside of, partially inside of, out side of, started by, finished by, contains of, covered by, covers, is- connected to, connects, is equal to, overlaps, is adjacent to, is disjunction from, intersects, includes, is included in, is- merged with } Spatial relations are mapped on the background knowledge on systems of qualitative spatial reasoning [18,19,20]. new relationships using architectural domain knowledge have been created as follows: S: {Distance, Orientation, Topologic: direct access with, indirect access with, overlooks, joints with} è Upper Level-type Relationships Definition: An upper-level type relationship, U 1, between 2 terms is defined as follows: U113: {hypernymy: target_term is a kind of term, hyponymy: term is a kind of target_term, holonymy: term1 is a part of term2, meronymy: parts of term, synonymy} These relationships have been borrowed from upper level ontologies and reasoning systems. Definition: An upper-level type relationship, U2, between 2 terms as components or entities is defined as follows: U2: {generalization: (entity) is a kind of
, specialization:
is a kind (entity), aggregation, part of 14} è Construction-type Relationships Definition: A construction-type relationship, C, between 2 terms as entities is defined by the following relations: C: { sticker of, coverer of, made of, fortifies, supports} The construction type relationship is the foundation of architecture that contains the techniques and process of building a physical structure with proper materials. Construction-type relationships cover the attributes of structure of a space. è Environment-type Relationships Definition: An environmental-type relationship, E, between 2 terms as entities is defined by the following relations: E: {cooler of, heater of, ventilated by, lighted by, decorated by} An architectural space always tries to create or enhance the quality of living environment. This characteristic produces environmentaltype relationships. 3.1.3. Designing Model of Relationships between Pair of Entities in Ontology Tool For defining the function of relationships between components in caravanserais, this research made a complementary analysis and synthesis of components not only in one case of caravanserais, but
As it is mentioned in definition of system, WordReference.com, French, German, Italian and Spanish Dictionary with Collins Dictionaries http://www.wordreference.com/ The best example of lexical database using the above relationships is Wordnet [7] Part of is discussed in Mereology which is theory of parts using whole/part relation as a substitute branch of logic http://plato.stanford.edu/entries/mereology/
Journal of Digital Information Management
q Volume 2 Number 4q December 2004
156
Figure 9 : Sketch of the graph of relationship links for the class of structural relationships.
in different samples, in order to cover the similarities and differences. Further each pair of terms as components or entities in the hierarchical structured term-set has been connected with a relationship, considering the direction of the link (starts from-ends to), and based on their characteristics in caravanserais. Fig 9 shows one of the several sketches of this research for defining relationships between components. As it has been mentioned, Protégé tool is used to design the above knowledge in a taxonomic model of classes, with instances and values or slots. Relationships are defined in Protégé in a hierarchy independent of terminology class. Each of the above groups of relationships is defined in classes and subclasses. Each component is attached to the class as an instance and is chosen from multilingual equivalent class. Fig 10 shows a snapshot of the model of relationships in protégé. As in most of the cases a relationship has a direction, 2 values are defined as slots from, and to, to connect related components. This model helps us to make the relationships languageindependent.
Journal of Digital Information Management
3.2. Issues on Defining, Designing and Implementing Relationships in Ontology for Caravanserais of Silk Roads The process of defining relationships in order to design a visual ontology model (in such a way that it covers the 3D architectural characteristics) is the most complicated task in this research. This process of ontology design tries to enrich the domain ontology capable enough to cover the attributes of architectural space. This process is a new approach in ontology development comparing with other ontologies that are designed in the field of cultural properties with application for visual data management, like ontologies for archeological objects and statues [21], art objects and furniture [22] and paintings[23]. Those researches focus on hierarchical ontology structures. Such a lack of background knowledge brings limitation as far as opportunities for our research. The initial prototype of spatial ontology has been created using sketches (as shown in Fig 9) and simple formula. Further, Protégé tool includes the structure for relationship support. upper-level relationships or those ones who will not be influenced by constraints,
q Volume 2 Number 4q December 2004
157
such as specialization of has been input in protégé tool. Meanwhile the corpus of historical architecture and caravanserais will bring constraints for this process of ontology design. Factors like historical period, or time, geographical location, function, construction founding, etc. will influence the function of defining relationships between 2 entities. Below examples are given for links that will differ bases on these constraints: -
Topology constraints: Room r is around of the yard Room
-
r
overlooks the yard (constraint: r first floor)
Typology constraints: Room r is around of caravanserais) Room r is around of caravanserais)
-
(constraint: r ground floor)
the yard (constraint: r desert the corridor (constraint: r mountain
Historical period constraints: Room r has direct access with caravanserais of 12 century)
the yard (constraint: r
Room r has indirect access with caravanserais of 17 century)
the yard (constraint: r
This research has been studying the possibilities of defining constraints between pair of entities especially in Protégé tool. Ontology Languages such as RDFS and OWL and the related interfaces is studies in order to define constraints for pair of links between instances. Meanwhile some mathematical properties such as reflexive, symmetric and transitive [24] can help the process of defining links and avoid repetition of some functions. Further discussion on the result of this research to fulfill the process of defining relationships for a spatial ontology with constraints and mathematical properties goes further than the framework of this paper. 3.3. Completing the Spatial- Lexical Ontology Model with Shape Rules The main challenge of the mentioned process of ontology design is to provide a model in which terms can describe spaces. There is always a gap between description of a space by words and what can be seen and described with shapes and finally what can be perceived in a space. The main question is how the values of a space, like form, circulation, geometry, structure, environment, etc. can be reflected in a multi-lingual lexical model. This task -which is a prototype now-, is an iterative process and needs to be extended after field research. We try to reduce the gap between formal specification of space and verbal descriptive model by designing spatial ontology. But there are still some formal values which terms can not fully describe it. For example form of a plan of a caravanserais and circulation inside this form can hardly be described by vocabularies. The key issue of this research is to find some other formal models for visualization of an architectural concept that use shapes, instead of vocabularies, to build a systematic visual model, such as the study of developing a shape grammar [25] for caravanserais. The future trend of this research combines formal rules to spatial ontology using shape grammar language.15
15
16 17
4. Conclusion The process of collecting or disseminating of visual data over internet, especially those which are dealing with concepts of spatial entity like architectural content, is becoming a challengeable task. The user might need to learn about technical content of such data, and computer should be able to recognize such content far from its ambiguity of shape, color, texture, form, etc. The main question which arises here is how the semantic of visual data can be extracted in such a way that computers also understand it and they use it to help for data acquisition. This article briefly represented the step toward solution of visual and architectural data management through designing a visual model as domain ontology and related technologies. The ontology for caravanserais of Silk Roads provides a visual knowledge-base so users as experts have a systematic access to lexical data of caravanserais, independent of visual changes of component in different perspectives or angles. It also provides opportunities to acquire technical knowledge about mutual relationships between parts (components) and wholes (spaces) in a caravanserai. One parallel related research on application it envisions, is the Image Learning Ontology System. This system, semi-automatically, creates semantic links between objects from images (recognizing and learning their shape) provided by the experts and technical terms from the Ontology on Silk Roads Caravanserais (Fig. 11). While the concept of ontology can visualize the knowledge related to architectural data in a lexical model, it is adopted to enable reuse of model by computers for more advanced IT applications. That means the ontology file produced by knowledge representation environments (e.g. Protégé) can be converted to machine understandable languages like RDF, RDFS and OWL. Based on it other tools or technologies (such as image learning ontology that is discussed above) can be developed or the computer agents can use it as a standard reference in the domain in their operations 16 [26]. All of this will help for a richer visual data representation available over internet for the end-users. Furthermore, it provides some progresses regarding the aim of future generation of internet, semantic web [27]. Finally, it can be emphasized that the visual lexical model developed for this typology of cultural heritage, the related technologies and its role in data representation over internet can also be considered as a systematic method for accessing records of historical relics or documentation of cultural heritage along spatial and temporal dimensions. We remind the reader that for some cases there are the only remaining data for future generation17. References [1]
T. R. Gruber, (1993). A translation approach to portable ontologies. Knowledge Acquisition, 5(2) 199-220.
See also: http://www-ksl.stanford.edu/kst/what-is-an-ontology.html [2]
N.F. Noy, Samson W. Tu. Developing Medical Informatics Ontologies with Protégé (Slides from the tutorial), http:// protege.stanford.edu/amia2003/index.html
[3]
W. Kleiss, M.Y. Kiani, (1995). Iranian caravanserais. Tehran :Iranian Cultural Heritage Organization. 794.
Shape grammars were invented in the end of the seventies by Stiny and Gips. They created an algorithm creating and understanding designs directly through computations with shapes, rather than indirectly through computations with text or symbols[25]. Further explanation related to this part of research goes beyond the subject of this paper For further information about the application of ontology for visual data management please refer to the references 21, 22, 23. The earthquake that has taken place in Bam in Iran in 26 December 2003 who ruined more that 80% of one of the greatest mud brick structure of the world, unique Citadel of Bam (Arg-e-Bam) can be considered as such examples of the heritage which we loose and are irreplaceable. Such disasters show the vital role of accurate documentation of cultural heritage
Journal of Digital Information Management
q Volume 2 Number 4q December 2004
158
Figure 10: A snapshot of relationship graph and an example of one instance of the subclass specialization
Figure 11: Sketch of the Image Learning Ontology tool for semantic annotation of the images.
Journal of Digital Information Management
q Volume 2 Number 4q December 2004
159
[4]
R. Hillenbrand, (2000). Islamic Architecture. Edinburgh: Edinburgh University Press.663.ISBN 0 7486 1379 X
[5]
T.R. Gruber, (1995). Toward principles for the design of ontologies used for knowledge sharing, International Journal of Human-Computer Studies, 43 (5-6) 907-928
[6]
N. Noy & D. L. McGuinness, (2001). Ontology Development 101: A Guide to Creating Your First Ontology. SMI Report Number: SMI-2001-0880
see also: http://protege.stanford.edu/publications/ ontology_development/ontology101.html [7]
Wordnet, a lexical database for English language http://www.cogsci.princeton.edu/~wn/
[8]
T. Peterson. Introduction to the Art and Architecture Thesaurus.(1994) Oxford University Press.
See also: http://www.getty.edu/research/conducting_research/ vocabularies/aat/about.html [10] A. Pease, I. Niles, and J. Li, (2002). The Suggested Upper Merged Ontology: A Large Ontology for the Semantic Web and its Applications. In Working Notes of the AAAI-2002 Workshop on Ontologies and the Semantic Web, Edmonton, Canada. See also: http://protege.stanford.edu/publications/ ontology_development/ontology101.html [11] N. Fridman Noy, R.W. Fergerson, and M.A. Musen, (2000), The knowledge model of protégé-2000: combining interoperability and flexibility, In knowledge Engineering and knowledge management: 12th International Conference EKAW2000, Juanles-Pins, volume 1937 of lecture notes in Artificial Intelligence, PP 17-32, Berline/Heidelberg. Also as: Technical Report Stanford University, School of Medicine, SMI-2000-0830 http://www-smi.stanford.edu/pubs/SMI_Reports/SMI-2000-0830.pdf [12] Database thesaurus of ministry for the Culture and Communication - direction of Architecture and InheritanceMérimée, http://www.culture.gouv.fr/public/mistral/thesarch_fr [13] E. Andaroodi, F. Andres, K. Ono, P. Lebigre, (2003). Ontology for caravanserais of Silk Roads: Needs, Processes, Constraints. Proceeding of Nara international symposium of Digital Silk Roads. 361-367. ISBN: 4-86049-024-x [14] GOLD, General Ontology for Linguistic Description http://emeld.douglass.arizona.edu:8080/index.html [15] ISO 5964, (1985). Documentation Guidelines for the Development and Establishment of Multilingual Thesauri
Journal of Digital Information Management
[17] A. Herskovits, (1987), Language and spatial cognition, Cambridge University Press, ISSN:0891-2017 [18] A.G. Cohn and S.M. Hazarika, (2001). Qualitative spatial representation and reasoning: An overview, Fundamentae Informaticae, No. 46, 2-32. [19] C. Freksa, R. Rohrig, (1993). Dimensions of qualitative spatial reasoning, In N. P. Carrete, M. G. Singh (Eds.), Qualitative reasoning and decision technologies, Proc. QUARDET93, 483 492. [20] Bonan Li, Guoray Cai, (2002). A general object-oriented spatial temporal data model, Proceeding of the symposium on Geospatial Theory, processing and applications, Ottawa, USA.
Visual Resources Association http://www.vraweb.org/vracore3.htm
[9]
[16] L. Talmy, (1983). spatial orientation: theory, research and application, chapter: How language structures space, New York: Plenum Press, 225 -282.
[21] Wun Soo, Chen-Yu Lee, Jaw Jium Yeh, Ching Chih Chen, (2002). Using Sharable Ontology to Retrieve Historical Images, International Conference on Digital Libraries, Proceedings of the second ACM/IEEE-CS joint conference on Digital libraries, Portland, Oregon, USA, 197-198 [22] B.J. Wielinga, A.Th Schreiber, J. Wielemaker, J.A.C. Sandberg, (2001). From Thesaurus to Ontology, International Conference on Knowledge Capture, Victoria, British Columbia, Canada.194201 [23] A. W. M. Smeulders, L. Hardman, G. Schreiber, and J. M. Geusebroek, (2002). An integrated multimedia approach to cultural heritage e-documents. In ACM Workshop on Multimedia Information Retrieval. ACM. [24] List of basic discrete mathematics topics http://en.wikipedia.org/wiki/ List_of_basic_discrete_mathematics_topics [25] T.Knight, Shape Grammar in Education and Practice: History and Prospect, http://www.mit.edu/~tknight/IJDC/ [26] A.Th. Schreiber, B. Dubbeldam, J.Weilemaker, and B.J. Wielinga. (2001). Ontology-based photo annotation,IEEE Intelligent Systems, May/June. 66-74 See also: http://www.cs.vu.nl/~guus/papers/Schreiber01a.pdf [27] T. Berners-Lee, Hendler-James, Lassila-Ora.(2001). The semantic web, a new web content that is meaningful to computers will unleash a revolution of new possibilities, Scientific American.com, Feature Article, May Issue See also: http://www.sciam.com/article.cfm?articleID=00048144-10D2-1C7084A9809EC588EF21&pageNumber=1&catID=2
q Volume 2 Number 4q December 2004
160
