Reference Ontology and (ONTO)2 Agent: The Ontology Yellow Pages

Please download to get full document.

View again

of 24
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Document Description
Reference Ontology and (ONTO)2 Agent: The Ontology Yellow Pages
Document Share
Document Tags
Document Transcript
  Reference Ontology and (ONTO) 2 Agent: The Ontology Yellow Pages Julio César Arpírez 1 , Asunción Gómez-Pérez 1 , Adolfo Lozano-Tello 2  and Helena Sofia Andrade N. P. Pinto 3 facultad de Informática, Universidad Politécnica de Madrid, Madrid, Spain 2 Departamento de Informática, Universidad de Extramadura, Extremadura, Spain 'Departamento de Engenharia Informática, Instituto Superior Técnico, Lisbon, Portugal Abstract.  Knowledge reuse by means of ontologies faces three important problems at present: (1) there are no standardized identifying features that characterize ontologies from the user point of view; (2) there are no web sites using the same logical organization, presenting relevant information about ontologies; and (3) the search for appropriate ontologies is hard, time-consuming and usually fruitless. To solve the above problems, we present: (1) a living set of features that allow us to characterize ontologies from the user point of view and have the same logical organization; (2) a living domain ontology about ontologies (called  Reference Ontology)  that gathers, describes and has links to existing ontologies; and (3) (ONTO) 2 Agent, the ontology-based WWW broker about ontologies that uses Reference Ontology as a source of its knowledge and retrieves descriptions of ontologies that satisfy a given set of constraints. 1.  Introduction and Motivation During recent years, considerable progress has been made in developing the conceptual bases for building technology that allows knowledge component reuse and sharing. One of the main motivations underlying both ontologies and problem-solving methods (PSM) is to enable sharing and reuse of knowledge and reasoning behavior across domains and tasks. PSMs and ontologies can be seen as complementary reusable components to construct knowledge systems (Gómez-Pérez  and Benjamins, 1998). Ontologies are concerned with static domain knowledge and PSMs with dynamic reasoning knowledge. The integration of ontologies and PSMs is a possible solution to the 'interaction problem' (Bylander and Chan-drasekaran, 1988), which states that representing knowledge for the purpose of solving some problem is strongly affected by the nature of the problem and the inference strategy to be applied to the problem. Ontologies are defined as a formal, explicit specification of a shared conceptualization (G. Gruber, 1993; Borst, 1997); that is, 'Conceptualization refers to an abstract model of some phenomenon in the world by having identified the relevant concepts of that phenomenon. Explicit means that the type of concepts used, and the constraints on their use are explicitly defined. Formal refers to the fact that the ontology should be machine-readable. Shared reflects the notion that an ontology captures consensual knowledge, that is, it is not private to some individual, but accepted by a group' (Studer et al, 1998). PSMs describe the reasoning process of a knowledge-based system in an implementation- and domain-independent manner (Benjamins and Fensel, 1998). There are also the notions of task ontologies (Mizoguchi et al, 1995) and PSM ontologies (Chandrasekaran et al, 1998). Nowadays, it is easy to get information from organizations that have ontologies and PSMs on the web. There are even accessible points that gather information about ontologies and have links to other web pages containing more explicit information about such ontologies (see The Ontology Page, also known as TOP; http :// and there are also ontology servers, like The Ontology Server ( (Farquhar et al, 1995, 1996), Cycorp's Upper CYC Ontology Server ( (Lenat, 1990) or Ontosaurus ( (Swartout et al, 1997), that collect a huge number of very well-known ontologies. In the PSM area, there are also many PSM repositories at different locations but they are not accessible for outsiders and they are not compatible (Benjamins et al, 1998). At present, the knowledge component reuse and sharing community has identified the need to provide intelligent agents or intelligent brokering services on the WWW that ease the search for such knowledge components. In the ontology field, the need for this kind of services was identified in Fikes and Farquhar (1997) and Foundation for Intelligent Physical Agents (1998), but there are no web sites that gather information about ontologies that have already been built using the same logical organization, and there are no intelligent brokers specializing in the ontology field that could help in this search. This paper presents (ONTO) 2 Agent, an ontology-based WWW broker that helps to select ontologies. There is a similar project in the PSM field, called IBROW3 (Benjamins et al, 1998), whose goal is to develop an intelligent brokering service for PSM reuse on the WWW. Apart from the problems arising from component search, the choice of a knowledge component that does not match the system needs properly, whose usage is expensive (people, hardware and software resources, time), or that has not been satisfactorily evaluated technically (verified and validated) may force future users to stop reusing the built knowledge component and oblige them to formalize the same knowledge again. These problems are probably what have led to there being relatively few applications known to date in areas like knowledge management, ontology-based brokers, natural language generation, enterprise modeling, knowledge-based systems, and interoperability between systems that reuse ontologies. This paper presents a solution to the problem of locating and searching the appropriate ontology on the web. End users usually face a complex multi-  criteria choice problem when they search for candidate ontologies. Apart from the dispersion of ontologies over several servers   (a) Ontology content formalization differs depending on the server at which it is stored. (b) Ontologies on the same server are usually described with different detail levels. (c) There is no common format for presenting relevant information about ontologies so that users can decide which ontology best suits their purpose. To speed up the use of ontologies in applications, our solution is to prepare a kind of  yellow pages of ontologies  that provide classified and updated information about ontologies following the same logical organization. These living 1  yellow pages help future users to locate candidate ontologies for a given application. (ONTO) 2 Agent, the  intelligent WWW broker  specialized in the ontology field that uses an ontology as its knowledge source, spreads information about existing ontologies, helps to search appropriate ontologies, reduces the search time for the desired ontology, and supplies pointers to the set of ontologies that totally/partially meet user requirements. This paper is organized as follows. In Section 2, we present an overview of our solution. Section 3 presents a set of features to characterize, compare, evaluate and assess ontologies from the user point of view. Section 4 shows how we have built the Reference Ontology (an ontology in the domain of ontologies) to be used by the broker specialized in the ontology field. Finally, Section 5 presents the OntoAgent architecture, that is, the technology we use to build ontology-based WWW brokers and how it has been instantiated in (ONTO) 2 Agent, the broker that answers questions in the domain of ontologies using the Reference Ontology as its knowledge source. (ONTO) 2  Agent is capable of answering questions about the features of ontologies that have been entered into the Reference Ontology. A specimen query would be: give me all the ontologies in the domain D that are implemented in languages LI and L2. 2.  An Overview Three main tasks were identified in our approach: 1. Characterization of domain knowledge, which, in our case, is the ontology field, by identifying a set of features that allows end users to compare different ontologies. 2.  Ontological Engineering, to build the ontology, which, in our case, is the Reference Ontology, a domain ontology in the domain of ontologies using the METHONTOLOGY framework (Fernández et al, 1997, 1999; Gómez-Pérez, 1996) and the Ontology Design Environment (Blázquez et al, 1998). 3.  Intelligent brokering services to access the ontology content, which, in our case, is the OntoAgent architecture, that is, the technology we have used to build ontology-based WWW brokers. Figure 1 shows how these tasks are related. The development of the ontology was divided into two phases. The first phase is centralized and concerns the Living in the sense that ontologies evolve and grow, as presented in Tennison and Shadbolt (1998).  Fig. 1. General overview of (ONTO) 2 Agent development process. development of the conceptual structure of the ontology, and the identification of its main concepts, taxonomies, relations, functions and axioms. The second phase is distributed among ontology developers and involves filling in WWW forms to enter knowledge about instances into the conceptual structure. The ontology (conceptual structure+instances) is centralized and stored in a relational database. Using a WWW interface, the user can consult the ontology using the domain ontology vocabulary. (ONTO) 2 Agent searches for instances that satisfy the query and provides the answer. 3.  Features for Comparing Ontologies 3.1.  Previous Work Although software engineering and knowledge engineering provide detailed features for evaluating the fitness of components and calculating their reusability in software applications (Slagle and Wick, 1988; Basili et al, 1994; Kan, 1995; Khairuddin and Key, 1995; Pressman, 1997), the literature reviewed in the field of ontologies shows that there are few papers on identifying features for describing, comparing and assessing ontologies. There are some articles about ontology features (e.g., Fridman and Hafner, 1997; Hovy, 1997; Arpirez, 1998; Uschold, 1998), and there are articles from which they can be extracted, although they do not specifically address ontology features (Uschold and Griininger, 1996). These papers are presented in chronological order below. One of the first articles from which we can extract general features to characterize ontologies is Uschold and Griininger (1996). The features identified are: formality, purpose and domain. Formality expresses the level of formality of one ontology. Its possible values are: informal, semi-informal, semi-formal and highly formal. Purpose characterizes the different uses of one ontology. The values  proposed are   communication, interoperability and systems engineering. Domain describes the piece of reality that the builder of the ontology wants to represent. The problem with this proposal is that it specifies only a few features and, hence, ontology characterization is limited. Another article presenting features for comparing ontologies is Fridman and Hafner (1997). In this proposal, features are classified as: general, related to the design process, related to the taxonomy, related to the internal concept structure and the relations between concepts, related to axioms and the inference mechanism, related to the applications built using the ontology, and important contributions. General features try to give an overview of the ontology; for instance, whether the ontology is general or domain-specific, its domain (if it is a domain-specific ontology), the size of the ontology, the formalism used to represent it, purpose, etc. Design process features characterize the building process and whether the ontology was evaluated. Taxonomy features characterize the structure of the ontology and the main concepts represented in the ontology. Internal structure and relations features characterize the concepts represented and the relations used to represent those concepts. Axioms and inference features characterize the kind of logic used to represent and infer knowledge. Application describes the applications built using the ontology. Contributions refer to the strengths and weaknesses of the ontology. This taxonomy of 25 features was used to compare CYC (Lenat, 1990), Wordnet (Miller, 1990), GUM (Bateman et al,  1994), Sowa's Ontology (Sowa, 1997), Dahlgren's Ontology (Dahlgren, 1988), UMLS (Humphreys and Lindberg, 1993), TOVE (Gruninger and Fox, 1995), GENSIM (Karp, 1993), Plinius (Van der Vet et al, 1994) and KIF (Genesereth and Fikes, 1992). The problems with this proposal are the limited characterization of ontologies, and the lack of criteria for classifying ontologies according to some of the features. Proposed at the same time as Fridman and Hafner (1997), Hovy (1997) presents a taxonomy of features for comparing ontologies for natural language processing. This proposal classifies its 36 features into form, content and us age.  Form features characterize the conceptual tools with which the ontology builder defines terms and axioms. They are related to mathematical logic and computational complexity. Some of the most important features according to this criterion are   parsimoniousness, type flipping and coercion, inheritance of proper ties,  support of inferences, worlds or contexts, TMS and non-first-order reasoning. Content features refer to terminology, axiom inferences and instances used by the ontology builder. They are related to philosophy, epistemology and the domain. Some of the most important features covered by this criterion are   size, organization, average specificity (general or domain-specific), linkage to other resources, characterization of axioms and inferences, instance fidelity and instance coverage. Finally, usage features describe the way in which the ontology builder builds, uses and maintains the ontology. They are related to software engineering and ergonomics. Some of the most important features covered by this criterion are   storage size, hardware and software platforms required, viewing and editing tools and documentation. The major weakness of this proposal is that it is incomplete. Although it covers a considerable number of features, some of the features are not defined, some are not clearly defined, some definitions contradict the examples and there are several features that are only relevant to natural language issues. Uschold (1998) proposes 10 features for classifying ontology applications: purpose, representation language and paradigms, meaning and formality, subject matter (domain), scale, development (is the ontology implemented, published,
Similar documents
View more...
Search Related
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks