Knowledge inference and retrieval

The protégé platform provides capabilities to integrate description logic reasoner engine, such as Jess, Clips, Racer, and so on. This study utilizes the Racer as a DL reasoner. The Racer provides several functionalities, such as consistency checking and taxonomy classifying for basic inference process. The consistency checking validates each class definition. If any inconsistent situation happened, the ontology has logic issue and can not be classified correctly. Designers are required to modify definitions of the ontology, such as asserted conditions. The taxonomy classifying is according to asserted conditions to generate an inferred hierarchy. In this hierarchy, it includes both asserted and inferred classes.

To retrieval knowledge from the developed ontologies, Protégé provides built-in knowledge acquisition (KA) functions by defining criteria in run time. Users also can develop their specific inference application by using related API, such as Jena API.

As illustrated in Figure 5.6, this study utilized Protégé query and KA functions to retrieval knowledge. The criteria are restricting in Rhododendron that had pink flowers, and erect steam. The response demonstrates that Dendrocharis is one of the qualified instances.

Since Protégé also can associate the digital contents in the Web, the related contents are also retrieved as illustrated in the bottom of Figure 5.6. The Web based knowledge retrieval system presented in this study was developed by Jena APIs (http://jena.sourceforge.net/) and is used as a knowledge reasoner. The reasoner loads both TBox and ABox when an inference process is triggered. Since the TBox contains terminological axioms (definitions), the reasoner confirms the consistency of assertions using corresponding definitions and calculates relations among assertions in the KBS Abox.

For example, in the knowledge retrieval process a user makes an inquiry regarding a

diagnostic description, such as a fever symptom. The system takes two treatment features of fever symptoms from the herbal drug ontology, "clearing away heat" and "purging intense heat", and generates a recommended herbal drug list. Figure 5.6 shows the outcome of the user selecting one of the herbal drugs. In this figure, the results of the first three sections are inference results that are implemented through the integration of both herbal drug and vascular plant ontologies. The first section provides general information regarding treatments and related herbal drugs. Meanwhile, the second section is abstract descriptions of the selected herbal drug. Moreover, the third section deals with the plant and its corresponding biological hierarchy tree. The herbal drug is generally represented using a common name rather than a scientific name. In this example, "Gardeniae Fructus" and "Gardenia Jasminoides Ellis" are the common and scientific names, respectively. The last section presents more descriptions of this plant, which are gathered from the NMNS content management system.

It is important to stress the key points in the inference process (see the expressions below this section). In herbal drug ontology, "Gardeniae Fructus" is used for the treatment "clearing away heat", that can be further inferred to be an appropriate corresponding herbal drug for treating fever symptoms. In the vascular plant ontology, "Gardenia Jasminoides Ellis" is a child of "Gardenia", and inherits the characteristic "clearing away heat" from "Gardenia."

Furthermore, the common name "Gardeniae Fructus" of the plant is "same as" the scientific name "Gardenia Jasminoides Ellis" in the logical manner. The process can be concluded based on the treatment details of fever symptom to find the logical "same as" relation between

"Gardeniae Fructus" and "Gardenia." Based on subsumption and "same as" relations

"Gardenia Jasminoides Ellis" can be identified as an appropriate plant for treating a fever symptom. All expressions are indirect relations between the symptoms of fever and "Gardenia Jasminoides Ellis." The power of the process is the inference capability that uses the implicit information to produce explicit knowledge.

5.7 Discussions and limitations

Figure 5.6 may look like a conventional Web page that can be published by any author who fills out proper data. However, the value of this study lies in demonstrating that the digital contents of NMNS can be promoted as knowledge sources, and can be reused by the public in the future. For example, suppose that if some parties have KBSs and wish to share their professional knowledge with others. Traditional techniques may only provide system integration rather than infer their contents in knowledge layers. That is, knowledge sharing involves not only system connection but also the participation of knowledge inference mechanisms. Particularly, this study has demonstrated that the OWL-DL can be used to construct an ontological KB. The OWL-DL supports describing concepts, attributes, and instances which contain a hierarchical description and logical relationship. Additionally, the

OWL is a machine readable language, which can automatically parse and share a knowledge base.

Jacob [43] has noted the difference between categorization and classification, which can be used to verify this study, particularly in creating an ontological KBS. Based on his explanation, categorization involves the process of dividing the world into groups of entities whose members share certain similarities. In the knowledge acquisition and representation phases of this study, FCA is used to collect common understandings of terminologies and their supporting attributes and relations as concepts. The OWL-DL then conducts detailed descriptions of each terminology that can be considered as categories (concepts). On the other hand, classification involves the orderly and systematic assignment of each entity to a single class within a system of mutually exclusive and non-overlapping classes. In the knowledge retrieval phase, the inference process analyzes terminology definition and then appropriately categorizes assertions.

The main limitations in this study are that the development of ontological KB is time-consuming and requires seamless collaboration among specialists, knowledge engineers, and information systems. Building an ontological KB is an art more than a technical skill.

Furthermore, a well-defined ontology requires long-term and constant maintenance. Therefore, this study proposes making ontologies smaller.

Chapter 6

Conclusion and future research

This study specifies various problems, including content silo trap problem among projects, content lost during activities, weakness throughout the entire knowledge management process, and poor provision of integrated services, existing in digital museum projects worldwide. A knowledge-based digital museum with curatorial, social, professional, administrative and promotional goals that uses unified knowledge content to provide proactive, adaptive, ubiquitous and collaborative services for various types of users through global knowledge management processes in actual and virtual accessible spaces is specified as the goal of the next generation digital museums.

To solve these problems and achieve the goal of creating a knowledge-based digital museum, this study proposes a knowledge-based digital museum framework and designs a practical system for application to the National Museum of Natural Science, Taiwan. The proposed framework attempts to solve overall knowledge management issues, including creation and management; reuse and diffusion; and sharing and exploration that can be fulfilled for a museum from both global and long-term perspectives. This framework is realized via the three real research approaches to demonstrate its practicability and benefits.

The details are discussed in Chapters 3, 4, and 5. This study yields three important results:

(1) Unified knowledge content management (UKCM) model

This model unifies abundant and growing knowledge content acquisition, representation, creation, organization, and publication for all specialists, departments, projects, applications, communities and domains based on common and standard workflow and multi-layer reusable knowledge content structures from both global and long-term perspectives.

(2) Ontological knowledge-based learning service model

This model reuses unified knowledge content to develop various value-added applications, particularly for learning services; and diffuses these contents and applications to various users proactively, adaptively and ubiquitously based common and sharable ontological knowledge concept.

(3) Ontological knowledge reuse and sharing approach

This approach shares knowledge content among domains, communities, and institutes and explores implicit and innovative knowledge in knowledge bases to expand coverage of knowledge domains using sharable and standard ontological techniques.

The above findings are in accord with the goals of the KBDM framework and satisfy the practical knowledge management requirements for NMNS. This study wishes to highlight some unique advantages of the proposed KBDM framework that are overlooked in global digital museum projects. These advantages, which are summarized as follows:

(1) Knowledge-based digital museum

This study constructs a knowledge-based digital museum to reveal the overall knowledge creation and management; knowledge reuse and diffusion; and knowledge sharing and exploration capabilities and impacts that most projects do not fulfill. First, knowledge workers create the contents and there is an authoritative basis for assuming them to be true. Second, the various contents are interpreted using metadata with semantic annotation created by knowledge workers. Third, ontological techniques represent all knowledge contents via a common knowledge concept.

(2) Unified knowledge content management approach

Some digital museum projects have applied knowledge management approaches but these projects have only considered dealing with problems involving partial

demonstrates the unified approach that knowledge management process is applied from the perspective of a global museum. Accordingly, the entire knowledge management process can be supported in issues involving creation and management, reuse and diffusion, and sharing and exploration

(3) Multi-layer reusable knowledge content structures

Most digital museum projects are focused on multimedia content or some other particular complex type of content, and thus their projects suffer from two major drawbacks. First, the limited content structure cannot fully support the contribution of domain know-how by knowledge workers. Second, the contents cannot efficiently support reuse in various services and applications. This study proposes multi-layer reusable knowledge content structures that knowledge workers can follow these structures to fully express their knowledge concept into systematic structures for users to capture them charily. The various content types can be flexibly organized and reused by various services and applications.

(4) Ubiquitous knowledge-based learning service model souvenirs

Museums are considered one the most important learning faculties in modern society. Numerous learning services focus on designing interactive user interfaces and multimedia presentations for specific applications with limited learning content.

Based on the unified knowledge bases, this study provides a learning service with abundant research, exhibition, and education knowledge content and integrated services, including support for ubiquitous, context-aware, personalized services.

(5) Ontological approach for entire knowledge management process

Numerous digital museum projects are aware that ontology is an effective tool for creating common and sharable knowledge concepts among domains, communities, and institutes. Few of these projects can apply ontological techniques covering knowledge creation, management, reuse, diffusion, sharing, and exploration

process to support applications such as this study.

Although this study’s findings have theoretical and practical implementations for constructing KBDM framework, this study is not problematic free. Some of the limitations of this study are highlighted below.

(1) Flexibility of knowledge content structures

It is uncertain whether the proposed multi-layer reusable knowledge content structures fully support the expression and organization of content for all possible applications. The problem is that it is difficult to predict what kind of structures will be needed in advance for a particular application. To date, we are certain that it is enough for expressing and organizing knowledge content for NMNS’s current applications. In the event of a future need to create new structures for certain applications, we believe that it will be easy to include them in the multi-layer structures.

(2) Completeness of sharable ontology

Designing ontology is tedious and time consuming. Building an ontological KB is more of an art than a technical skill. The greatest challenge is the process of obtaining consensus among individuals, domains, and communities. Therefore, this study simply presents the methodology for constructing ontologies via an example involving two related domains. This approach enables the gradual construction of the ontologies among correlated domains and communities.

(3) Development of user segmentation content

Content creation is costly in terms of time and money as well as being labor intensive. It is not practical to create different content for different groups of users. It is better to create content for the major user groups and applications within limited resources. Currently, this study simply focuses on creating content for major groups

(4) Balance between systematic generation and art design content

The system generates a vast amount of knowledge content. This significantly reduces the burden faced by system engineers in converting this content into web pages. However, the visual quality and art creativity are constrained due to the limited presentation templates used. Accordingly, it is necessary to compromise between efficiency and visual quality for individual applications. If visual presentation is emphasized, supplemental art design is necessary following system generation processing.

(5) Decision maker support

Although this study proposes an entire knowledge management framework and realizes it via practical approaches, the content must still be created by knowledge specialists. Considerable knowledge content must also be delivered through various applications and promoted through various intelligent services. Besides enormous resource support, particularly in terms of techniques, manpower and funds, the main success factor is support from decision makers with global and long-term vision.

Consequently, numerous contents, applications and services can be created continuously without barriers via top-down strategy.

Future research is clearly required, but this study is a crucial step. Among the numerous topics to be investigated in future researches, some important ones are as follows:

(1) Intelligent knowledge discovery and exploration mechanism

Research on knowledge discovery, classification, and organization is underway.

Currently, a classification hierarchy is being used; a partial ontologies via an example involving two related domains is constructed, and in the future this hierarchy will be developed into a fully-fledged ontology among domains. The innovative implicit knowledge amassed in core and advanced knowledge elements will be generated and inferred inside the same domain or among related domains automatically. Moreover,

the ontology will be dynamically extended, maintained, and mapped between specialists and users during content creation. Figure 6.1 illustrates the ontological knowledge discovery framework.

Figure 6.1 Ontological knowledge discovery framework

(2) Ontology-based personalized service model

The ubiquitous learning service model will be extended to all applications to construct a ubiquitous knowledge-based digital museum portal. An advanced personalization service for serving various users and applications will shortly be created. A more intelligent web mining personalization service that includes a personal ontology and natural language query will be created based on the improvement of ontology and data mining techniques. Figure 6.2 shows the personalization service model to support all applications and every individual user

Figure 6.2 Personalization service for ubiquitous knowledge-based digital museum (3) User behavior and satisfaction evaluation

This study creates extensive unified knowledge content based on multi-layer reusable knowledge content structures, and used these structures to develop ubiquitous, context-aware, and personalized learning services for users. However, this study further needs to know whether the contents and services can meet user requirements and whether the proposed system can achieve good performance for each service. Therefore, it is necessary to design evaluation methods by monitoring service response from the system and relevance feedback from users. Such evaluations can help us to improve and adjust content quantity and quality and service accuracy and efficiency.

(4) Advanced online authoring tool for designing applications

The proposed system provides multi-layer reusable knowledge content structures

with systematic authoring tools for organizing various types of contents for various applications. These authoring tools can easily construct knowledge elements using predefined structures. However, more visual and flexible systematic authoring tools are required to support curators and teachers to design digital exhibition or educational material online by utilizing the vast existing knowledge bases.

(5) Knowledge-based digital museum business model

The KBDM framework has significantly extended the curatorial, social, professional, administrative and promotional goals of physical museums. This study hopes that this framework can incorporate with e-commerce model to serve knowledge consumers and digital industry. Therefore, the current framework should be extended to support commercial operation and promotion for both domestic and international considerations. This new generation digital museum business model can support museums not only for fulfilling original functions but also for creating revenue in the current knowledge economy era. Figure 6.3 illustrates the ubiquitous knowledge-based digital museum business model.

Figure. 6.3 Ubiquitous knowledge-based digital museum business model

Finally, this study stresses again that its objective is not to solve all knowledge management issues for all museums. On the other hand, this study tries to develop a knowledge management methodology that can provide a worthy reference for museums that possess vast quantities of knowledge. This framework can not only achieve unified knowledge management problems but also create unlimited opportunities and competition based on the vast knowledge content and numerous innovative services possessed by museums.

Bibliography

[1] Ackoff, R. L., “From Data to Wisdom,” Journal of Applied Systems Analysis, no. 16, pp.

3-9, 1989.

[2] Alchin, N., Theory of knowledge, Hodder Murray, London, 2003.

[3] Addison, A. C., ”Virtual heritage, archeology, and cultural heritage,” Proceedings of the 2001 Conference on Virtual Reality, Archeology, and Cultural Heritage, pp. 343-354, Glyfada, Greece, November 2001.

[4] Amann, B., Fundulaki, I., and Scholl, M., “Integrating ontologies and thesauri for RDF schema creation and metadata querying,” International Journal on Digital Libraries Studies, vol. 3, no. 3, pp. 221, 2000.

[5] Anani, N., “Enhancing The Heritage Experience,” Museum and the Web 2005, Vancouver, BC, Canada, April 2005.

[6] Baader, F., Calvanese, D., McGuinness, D., Nardi, D., and Parel-Schneider, P.(eds), The Description Logic Handbook, Cambridge University Press, Cambridge, UK, January 2003.

[7] Batini, C., Ceri, S., and Navathe, S. B., Conceptual database design: An entity-relationship approach, The Benjamin/Cummings Publishing Company Inc., Redwood City, California, 1992.

[8] Bellinger, G., “Knowledge Management—Emerging Perspectives,” 2004, available at http://www.systems-thinking.org/ (accessed 3 March 2006).

[9] Boiko, B., Content Management Bible, Wiley Publishing, Inc, New York, NY, 2002.

[10] Bonett, M., ”Personalization of Web Services: Opportunities and Challenges,” Ariadne, iss. 28, June 2001, available at http://www.ariadne.ac.uk/issue28/personalization/

[11] Cannon-Brookes, P., “The Nature of Museum Collections,” Manual of Ccuratorship, 2nd.

ed. Ed. J. Thompson., pp. 500-512, London, 1992.

[12] Chao, K., Smith, P., Hills, W., Florida-James, B., and Norman, P., “Knowledge sharing are reuse for engineering design integration,” Expert Systems with Applications, vol.

14, no. 3, pp. 399-408, 1998.

[13] Chen, B. H., Hung, S. H., and Hong, J. S., “Modularization framework for digital museum exhibition,” Proceedings of the third ACM/IEEE-CS joint conference on Digital libraries, pp. 394-394, Houston, Texas, May 2003.

[14] Chen, P. P., “The entity-relationship model toward a unified view of data,” ACM Transaction on Database Systems, vol. 1, no 1, pp. 9-36, 1976.

[15] Cinotti, T. S., Raffa, G., and Roffia, L., “Evaluating Context-Aware Mobile Applications in Museums: Experience from the MUSE Project,” Museum and the Web 2004 conference, Arlington, Virginia / Washington DC, April 2004, available at http://www.archimuse.com/mw2004/papers/salmon/salmon.html (accessed 20 September 2005).

[16] Common Information Environment group, “Understanding the Audience,” February 2005, available at http://www.common-info.org.uk/docs/mori-report.pdf (accessed 20 March 2006).

[17] Dai, H. H. and Mobasher, B., “A Road map to More Efffective Web personalization:

Integrating Domain Knowledge with web Usage mining,” Proceedings of the International Conference on Internet Computing 2003, Las Vegas, Nevada, USA, June 2003.

[18] Davenport, T. H. and Prusak, L., Working knowledge: How organizations manage what they know, Boston, MA: Harvard Business School Press, 1998.

[19] Decker, S., Melnik, S., Harmelen, F. V., Fensel, D., Klein, M., Broekstra, J., Erdmann, M., and Horrocks, I., “The Semantic Web: The Roles of XML and RDF,” Internet

[19] Decker, S., Melnik, S., Harmelen, F. V., Fensel, D., Klein, M., Broekstra, J., Erdmann, M., and Horrocks, I., “The Semantic Web: The Roles of XML and RDF,” Internet