Gerstner Laboratory for Intelligent Decision Making and Control Czech Technical University in Prague -
Series of Research Reports Report No:
GL 126/01
Ontologies Description and Applications Marek Obitko
[email protected] http://cyber.felk.cvut.cz/gerstner/reports/GL126.pdf
Gerstner Laboratory, Department of Cybernetics Faculty of Electrical Engineering, Czech Technical University Technická 2, 166 27 Prague 6, Czech Republic tel. (+420-2) 2435 7421, fax: (+420-2) 2492 3677 http://cyber.felk.cvut.cz/gerstner
Prague, 2001 ISSN 1213-3000
Ontologies – Description and Applications Marek Obitko
[email protected] February 22, 2001
Abstract The word “ontology” has gained a good popularity within the AI community. Ontology is usually viewed as a high-level description consisting of concepts that organize the upper parts of the knowledge base. However, meaning of the term “ontology” tends to be a bit vague, as the term is used in different ways. In this paper we will attempt to clarify the meaning of the ontology including the philosophical views and show why ontologies are useful and important. We will give an overview of ontology structures in several particular systems. A field proposed within ontological efforts, “ontological engineering”, will be also described. Usage of ontologies in several particular ways will be discussed. These include systems and ideas to support knowledge base sharing and reuse, both for computers and humans, ontology based communication in multiagent systems, applications of ontologies for natural language processing, applications in documents search and enrichment of knowledge bases, both particularly for the World Wide Web environment and construction of educational systems, particularly intelligent tutoring systems.
1
Contents 1 Introduction 1.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.2 Philosophical View . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3 3 4
2 What is an Ontology? 2.1 Common Definitions . . . . . . . . . . . . . . . . . . . . 2.1.1 Ontology as a Philosophical Term . . . . . . . . 2.1.2 Ontology as a Specification of Conceptualization 2.1.3 Ontology as a Representational Vocabulary . . . 2.1.4 Ontology as a Body of Knowledge . . . . . . . . 2.2 Other Ontology Definitions . . . . . . . . . . . . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
4 4 5 5 6 7 8
3 Ontology Structure 3.1 CYC . . . . . . . . . . . . . . . . . . . 3.2 Russell & Norvig’s General Ontology . 3.3 Ontology Engineering . . . . . . . . . 3.3.1 Structure of Usage . . . . . . . 3.3.2 Ontology Engineering Subfields
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
9 9 10 11 11 12
4 Using Ontologies 4.1 Top-Level Ontology . . . . . . . . . . . . . . . . . 4.2 Knowledge Sharing and Reuse . . . . . . . . . . . . 4.2.1 KIF . . . . . . . . . . . . . . . . . . . . . . 4.2.2 Ontolingua . . . . . . . . . . . . . . . . . . 4.2.3 Collaboration . . . . . . . . . . . . . . . . . 4.2.4 Particular Ontologies Reuse . . . . . . . . . 4.3 Communication in Multi-Agent Systems . . . . . . 4.3.1 FIPA Agent Management Model . . . . . . 4.3.2 Ontology Service by FIPA . . . . . . . . . . 4.3.3 Ontologies Relationships . . . . . . . . . . . 4.3.4 FIPA Knowledge Model and Meta-Ontology 4.4 Natural Language Understanding . . . . . . . . . . 4.4.1 CYC NLP . . . . . . . . . . . . . . . . . . . 4.4.2 WordNet Database . . . . . . . . . . . . . . 4.5 Document Search and Ontologies . . . . . . . . . . 4.5.1 OntoSeek . . . . . . . . . . . . . . . . . . . 4.5.2 WebKB . . . . . . . . . . . . . . . . . . . . 4.5.3 Knowledge Representation Techniques . . . 4.5.4 Document Enrichment . . . . . . . . . . . . 4.6 Educational Systems . . . . . . . . . . . . . . . . . 4.6.1 EON . . . . . . . . . . . . . . . . . . . . . . 4.6.2 ABITS . . . . . . . . . . . . . . . . . . . . . 4.6.3 Other Proposals . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . .
14 15 16 16 16 17 18 18 19 20 21 22 23 24 24 24 25 26 26 27 29 29 30 30
5 Conclusion
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
30
2
1
Introduction
The word “ontology” has gained a good popularity within the AI community. Ontology is usually viewed as a high-level description consisting of concepts that organize the upper parts of the knowledge base. However, meaning of the term “ontology” tends to be a bit vague, as the term is used in different ways. In this paper we will attempt to clarify the meaning of the ontology and show why ontologies are useful and important. We will discuss usage of ontologies in several particular ways, such as knowledge base reuse, knowledge sharing, communication in multi-agent systems, applications of ontologies for WWW applications, for natural language processing, and for intelligent tutoring systems.
1.1
Motivation
In AI research history, we can identify two types of research [31, 8]. One is form-oriented research (mechanism theories) and the other is content-oriented research (content theories). The former deals with logic and knowledge representation while the latter with content of knowledge. Apparently, the former has dominated AI research up to date. Recently, however, content-oriented research has become to gather much attention because a lot of real-world problems to solve such as knowledge reuse, facilitation of agent communication, media integration through understanding, large-scale knowledge bases, and so on, require not only advanced theories or reasoning methods, but also sophisticated treatment of the content of knowledge. Formal theories such as predicate logic provides us with a powerful tool to guarantee sound reasoning and thinking. It even enables us to discuss the limits of our reasoning in a principled way. However, it cannot answer to any of the questions such as what knowledge we should have for solving given problems, what is knowledge at all, what properties a specific knowledge has, and so on. Sometimes, the AI community gets excited by some mechanisms such as neural nets, fuzzy logic, genetic algorithms, constraint propagation etc. These mechanisms are proposed as the “secret” of making intelligent machines. At other times, it is realized that, however wonderful the mechanism, it cannot do much without a good content theory of the domain on which it is to work. Moreover, we often recognize that once a good content theory is available, many different mechanisms might be used equally well to implement effective systems, all using essentially the same content. Importance of content-oriented research is being recognized more and more nowadays. Unfortunately it seems that there are no widely recognized sophisticated methodologies for content-oriented research now. Major results till later years were only development of knowledge bases. The reasons for this can be [31]: • content-oriented research tends to be ad-hoc • there is no methodology that enables to accumulate research results It is necessary to overcome these difficulties in the content-oriented research. Ontologies are proposed for that purpose. Ontology engineering, as proposed in e.g. [31], is a research methodology which gives us design rationale of a knowledge base, kernel conceptualization of the world of interest, strict definition of basic meanings of basic concepts together with sophisticated theories and technologies enabling accumulation of knowledge which is dispensable for modeling the real world. 3
Interest in ontologies has also grown as researchers and system developers have become more interested in reusing or sharing knowledge across systems. Currently, one key impediment to sharing knowledge is that different systems use different concepts and terms for describing domains. These differences make it difficult to take knowledge out of one system and use it in another. If we could develop ontologies that could be used as the basis for multiple systems, they would share a common terminology that would facilitate sharing and reuse. Developing such reusable ontologies is an important goal of ontology research. Similarly, if we could develop tools that would support merging ontologies and translating between them, sharing would be possible even between systems based on different ontologies.
1.2
Philosophical View
The term ontology was taken from philosophy. According to Webster’s Dictionary an ontology is • a branch of metaphysics relating to the nature and relations of being • a particular theory about the nature of being or the kinds of existence Ontology (the “science of being”) is a word, like metaphysics, that is used in many different senses. It is sometimes considered to be identical to metaphysics, but we prefer to use it in a more specific sense, as that part of metaphysics that specifies the most fundamental categories of existence, the elementary substances or structures out of which the world is made. Ontology will thus analyze the most general and abstract concepts or distinctions that underlay every more specific description of any phenomenon in the world, e.g. time, space, matter, process, cause and effect, system. Recently, the term of “ontology” has been up taken by researchers in Artificial Intelligence, who use it to designate the building blocks out of which models of the world are made. An agent (e.g. an autonomous robot) using a particular model will only be able to perceive that part of the world that his ontology is able to represent. In this sense, only the things in his ontology can exist for that agent. In that way, an ontology becomes the basic level of a knowledge representation scheme. An example is set of link types for a semantic network representation which is based on a set of ”ontological” distinctions: changing–invariant, and general–specific.
2
What is an Ontology?
The term “ontology” is used in many different ways. In this section we will discuss what an ontology is on several definitions that are currently used.
2.1
Common Definitions
The most widespread definitions of ontology are given below. 1. Ontology is a term in philosophy and its meaning is “theory of existence”. 2. Ontology is an explicit specification of conceptualization [21]. 3. Ontology is a theory of vocabulary or concepts used for building artificial systems [31]. 4
4. Ontology is a body of knowledge describing some domain (eg. a common sense knowledge domain in CYC [45]) The definition 1 is radically different from all the others (including additional ones discussed below). We will shortly discuss some implications of its meaning for definition of “ontology” for AI purposes. The second definition is generally proposed as a definition of what an ontology is for the AI community. It may be classified as “syntactic”, but its precise meaning depends on the understanding of the terms “specification” and “conceptualization”. The third definition is a proposal for definition within the knowledge engineering community. The last fourth definition differs from the previous two ones — it views the ontology as an inner body of knowledge, not as the way to describe the knowledge. Although these definitions are compact, they are not sufficient for in-depth understanding of what an ontology is. We will try to give more comprehensive definitions and insights. 2.1.1
Ontology as a Philosophical Term
Following [24] we will use the convention that the uppercase initial letter “O” is to distinguish the “Ontology” as a philosophical discipline from other usages of this term. Ontology is a branch of philosophy that deals with the nature and the organization of reality. It tries to answer questions like “what is existence”, “what properties can explain the existence” etc. Aristotle defined Ontology as the science of being as such. Unlike the special sciences, each of which investigates a class of beings and their determinations, Ontology regards “all the species qua being and the attributes that belong to it qua being” (Aristotle, Metaphysics, IV, 1). In this sense Ontology tries to answer the question “what is the being?” or, in a meaningful reformulation “what are the features common to all beings?”. This is what is today called “General Ontology” in contrast with various Special or Regional Ontologies (eg. Biological, Social). From this, Formal Ontology is defined as an area that has to determinate the conditions of the possibility of the object in general and the individualization of the requirements that every object’s constitution has to satisfy. According to [24] Formal Ontology can be defined as the systematic, formal, axiomatic development of the logic of all forms and modes of being. From this, Formal Ontology is not concerned so much in the existence of certain objects, but rather in the rigorous description of their forms of being, i.e. their structural features. In practice, Formal Ontology can be intended as the theory of the distinctions, which can be applied independently of the state of the world, i. e. the distinctions: • among the entities of the world (physical objects, events, regions...) • among the meta-level categories used to model the world (concept, property, quality, state, role, part...) In this sense, Formal Ontology, as a discipline, may be relevant to both Knowledge Representation and Knowledge Acquisition [24]. 2.1.2
Ontology as a Specification of Conceptualization
The second definition of ontology mentioned above, explicit specification of conceptualization, is briefly described in [20]. The definition comes from work [22] where the ontology is 5
used in context of knowledge sharing. According to Thomas Gruber, explicit specification of conceptualization means that an ontology is a description (like a formal specification of a program) of the concepts and relationships that can exist for an agent or a community of agents. This definition is consistent with the usage of ontology as set of concept definitions, but more general. In this sense, ontology is important for the purpose of enabling knowledge sharing and reuse. An ontology is in this context a specification used for making ontological commitments. Practically, an ontological commitment is an agreement to use a vocabulary (i.e. ask queries and make assertions) in way that is consistent (but not complete) with respect to the theory specified by an ontology. Agents are then built that commit to ontologies and ontologies are designed so that the knowledge can be shared with and among these agents. The body of a knowledge is based on a conceptualization: the objects, concepts, and other entities that are assumed to exist in some area of interest and the relationship that hold among them. A conceptualization is an abstract, simplified view of the world that we wish to represent for some purpose. Every knowledge base, knowledge-based system, or knowledge-level agent is committed to some conceptualization, explicitly or implicitly. For these systems, what “exists” is that which can be represented. When the knowledge of a domain is represented in a declarative formalism, the set of objects that can be represented is called the universe of discourse. This set of objects and the describable relationships among them, are reflected in the representational vocabulary with which a knowledge-based program represents knowledge. Thus, in the context of AI, we can describe the ontology of a program by defining a set of representational terms. In such an ontology, definitions associate the names of entities in the universe of discourse (e.g. classes, relations, functions, or other objects) with human readable text describing what the names mean, and formal axioms that constraint the interpretation and well-formed use of these terms. Formally it can be said that an ontology is a statement of a logical theory [20]. Ontologies are often equated with taxonomic hierarchies of classes without class definitions and the subsumption relation. Ontologies need not to be limited to these forms. Ontologies are also not limited to conservative definitions, that is, definitions in the traditional logic sense that only introduce terminology and do not add any knowledge about the world. To specify a conceptualization, one needs to state axioms that do constrain the possible interpretations for the defined terms. Pragmatically, a common ontology defines the vocabulary with which queries and assertions are exchanged among agents. The agents sharing a vocabulary need not share a knowledge base. An agent that commits to an ontology is not required to answer all queries that can be formulated in the shared vocabulary. In short, a commitment to a common ontology is a guarantee of consistency, but not completeness, with respect to queries and assertions using the vocabulary defined in the ontology. 2.1.3
Ontology as a Representational Vocabulary
The third definition of ontology proposed above says that it is in fact a representational vocabulary [8, 31]. The vocabulary can be specialized to some domain or subject matter. More precisely, it is not the vocabulary as such that qualifies as an ontology, but the conceptualization that the terms in the vocabulary are intended to capture. Thus, translating the terms in an ontology from one language to another, for example from Czech to English, does not change the ontology conceptually. In engineering design, one might discuss the ontology 6
of an electronic devices domain, which might include vocabulary that describes conceptual elements — transistors, operational amplifiers, and voltages — and the relations between these elements — operational amplifiers are a type-of electronic device, and transistors are component-of operational amplifiers. Identifying such a vocabulary and the underlying conceptualization generally requires careful analysis of the kinds of objects and relations that can exist in the domain. The term ontology is sometimes used to refer to a body of knowledge describing some domain (see below), typically a common sense knowledge domain, using a representational vocabulary. For example, CYC [45] often refers to its knowledge representation of some area of knowledge as its ontology. In other words, the representation vocabulary provides a set of terms with which one can describe the facts in some domain, while the body of knowledge using that vocabulary is a collection of facts about a domain. However, this distinction is not as clear as it might first appear. In the electronic-device example, that transistor is a component-of operational amplifier or that the latter is a type-of electronic device is just as much a fact about its domain as a CYC fact about some aspect of space, time or numbers. The distinction is that the former emphasizes the use of ontology as a set of terms for representing specific facts in an instance of the domain, while the latter emphasizes the view of ontology as a general set of facts to be shared. 2.1.4
Ontology as a Body of Knowledge
Sometimes, ontology is defined as a body of knowledge describing some domain, typically a common sense knowledge domain, using a representation vocabulary as described above. In this case, an ontology is not only the vocabulary, but the whole “upper” knowledge base (including the vocabulary that is used to describe this knowledge base). The typical example of this definition usage is project CYC (http://www.cyc.com/, [45]) that defines its knowledge base as an ontology for any other knowledge based system. CYC is the name of a very large, multi-contextual knowledge base and inference engine. The development of CYC started during the early 1980s headed by Douglas Lenat. CYC is an attempt to do symbolic AI on a massive scale. It is neither based on numerical methods such as statistical probabilities, nor is it based on neural networks or fuzzy logic. All of the knowledge in CYC is represented declaratively in the form of logical assertions. CYC contains over 400, 000 significant assertions [45], which include simple statements of fact, rules about what conclusions to draw if certain statements of fact are satisfied (true), and rules about how to reason with certain types of facts and rules. New conclusions are derived by the inference engine using deductive reasoning. The CYC team doesn’t believe there is any shortcut toward being intelligent or creating an artificial intelligence based agent. Addressing the need for a large body of knowledge with content and context may only be done by manually organizing and collating information. This knowledge includes heuristic, rule of thumb problem solving strategies, as well as facts that can only be known to a machine if it is told. Much of the useful common sense knowledge needed for life is prescientific and has therefore not been analyzed in detail. Thus a large part of the work of the CYC project is to formalize common relationships and fill in the gaps between the highly systematized knowledge used by specialists. It is not necessary to divide such a large knowledge base into smaller pieces to enable 7
reasoning in reasonable time. Because of this, the CYC knowledge base uses a special context space [29], that is divided by 12 dimensions into smaller pieces (contexts) that have something in common and can be used to reason about a specific problem in that context. It is possible to “lift” assertion from one context to another when the problem requires it. The CYC common sense knowledge can be used as a body of a knowledge base for any knowledge intensive system. In this sense, this body of knowledge can be viewed as an ontology of the knowledge base of the system.
2.2
Other Ontology Definitions
As we can see from the above discussions, the exact definition of ontology is not obvious, however it can be seen that the definitions have much in common. In addition to the above definitions there are many other proposals for ontology definitions. Some other definitions collected from [24] are: 1. informal conceptual system 2. formal semantic account 3. representation of a conceptual system via a logical theory (a) characterized by specific formal properties (b) characterized only by its specific purposes 4. vocabulary used by a logical theory 5. (meta-level) specification of a logical theory Definitions 1 and 2 conceive an ontology as a conceptual “semantic” entity, either formal or informal, while according to the interpretations 3, 4 and 5 is a specific “syntactic” object. According to interpretation 1, an ontology is the conceptual system which may be assumed to underlay a particular knowledge base. Under interpretation 2, instead, the ontology, that underlies a knowledge base, is expressed in terms of suitable formal structures at the semantic level. In both cases, we may say that “the ontology of knowledge base A is different from that of knowledge base B”. Under interpretation 3, an ontology is nothing else then a logical theory. The issue is whether such a theory needs to have particular formal properties in order to be an ontology or, rather, whether it is the intended purpose which lets us consider a logical theory as an ontology. The latter position can be supported by arguing that an ontology is an annotated and indexed set of assertion about something: “leaving off the annotations and indexing, this is a collection of assertions: what in logic is called a theory” (Pat Hayes statement in [24]). According to interpretation 4, an ontology is not viewed as a logical theory, but just as the vocabulary used by a logical theory. Such an interpretation collapses into 3.a if an ontology is thought of as a specification of a vocabulary consisting of a set of logical definitions. We may anticipate that the Gruber’s interpretation (specification of conceptualization) collapses into 3.a as well when a conceptualization is intended as a vocabulary. Finally, under interpretation 5, an ontology is seen as a specification of a logical theory in the sense that it specifies the “architectural components” (or primitives) used within a particular domain theory. 8
3
Ontology Structure
From the overview above we can see that an ontology can be perceived in basically two approaches. The first approach is an ontology as a representational vocabulary, where the conceptual structure of terms should remain unchanged during translation. The other approach, that is discussed in this section, is an ontology as the body of knowledge describing a domain, in particular a common sense domain. An ontology can be divided in several ways. We will describe some of the proposals here. Particularly interesting is so called “upper ontology” that is intended to serve as an upper part of ontology of practically all knowledge based systems. Some of the ways of dividing presented here are intended to be used for merging to form an upper ontology standard in the IEEE Standard Upper Ontology Study Group [39]. On pages linked from [39] there are many other examples that could be used as some kind of an upper ontology. CYC
Individual object
Wordnet
Thing
Intangible
Represented
Living
GUM
Nonliving
Sowa's Um-Thing
Configuration
Thing
Element
Thing
Sequence
Concrete
Process
Object
Abstract
Figure 1: How ontologies differ in their analyses of the most general concepts [8] It is interesting that many authors agree that the upper class1 of the ontology is “thing”, however even in the second level they do not agree on the separation, as can be seen in the figure 1. The initiative [39] tries to unify these views.
3.1
CYC
The ontology of CYC is based on a several terms that form the fundamental vocabulary of the CYC knowledge base. The universal set is #$Thing2 (see figure 1). It is the set of everything. Every CYC constant in the knowledge base is a member of this collection. In the prefix notation of the language CycL [10], we express that fact as (#$isa CONST #$Thing). Thus, too, every collection in the knowledge base is a subset of the collection #$Thing. In CycL, that fact is expressed as (#$genls COL #$Thing). The set #$Thing has some subsets, such as PathGeneric, Intangible, Individual, SimpleSegmentOfPath, PathSimple, MathematicalOrComputationalThing, IntangibleIndividual, Product, TemporalThing, SpatialThing, Situation, EdgeOnObject, FlowPath, Computation1 When using hierarchy of classes for ontology description. For describing ontologies however we do not have to limit to class hierarchies as in the case of taxonomies. 2 We will use the notation used in CYC language. The explanation of it can be found in [10].
9
alObject, Microtheory, plus about 1500 more public subsets and about 13600 unpublished subsets. #$Individual is the collection of all things that are not sets or collections. Thus, #$Individual includes (among other things) physical objects, temporal subabstractions of physical objects, numbers, relations, and groups (#$Group). An element of #$Individual may have parts or a structure (including parts that are discontinuous), but no instance of #$Individual can have elements or subsets. #$Collection is the collection of all CYC collections. CYC collections are natural kinds or classes, as opposed to mathematical sets. Their elements have some common attribute(s). Each CYC collection is like a set in so far as it may have elements, subsets, and supersets, and may not have parts or spatial or temporal properties. Sets, however, differ from collections in that a mathematical set may be an arbitrary set of things which have nothing in common (#$Set-Mathematical). In contrast, the elements of a collection will all have in common some feature(s), some ‘intensional’ qualities. In addition, two instances of #$Collection can be co-extensional (i.e. have all the same elements) without being identical, whereas if two arbitrary sets had the same elements, they would be considered equal. #$Individual and #$Collection are disjoint collections. No CYC constant can be an instance of both. #$Predicate is the set of all CYC predicates. Each element of #$Predicate is a truthfunctional relationship in CYC which takes some number of arguments. Each of those arguments must be of some particular type. Informally, one can think of elements of #$Predicate as functions that always return either true or false. More formally, when an element of #$Predicate is applied to the legal number and type of arguments, an expression is formed which is a well-formed formula (wff) in CycL. Such expressions are called atomic formulas if they contain variables, or ground atomic formulas (gaf) if they contain no variables. #$isa: expresses the ISA relationship. (#$isa EL COL) means that EL is an element of the collection COL. CYC knows that #$isa distributes over #$genls. That is, if one asserts (#$isa EL COL) and (#$genls COL SUPER), CYC will infer that (#$isa EL SUPER). Therefore, in practice one only manually asserts a small fraction of the #$isa assertions — the vast majority are inferred automatically by CYC. #$genls: expresses similar relationship for collections (generalization). (#$genls COL SUPER) means that SUPER is one of the supersets of COL. Both arguments must be elements of #$Collection. Again, as with the #$isa, CYC knows that #$genls is transitive, therefore, in practice one only manually asserts a small fraction of the #$genls assertions since the rest is inferred inferred automatically. More details about the structure of the CYC ontology and about how the CYC knowledge base is constructed can be found at http://www.cyc.com.
3.2
Russell & Norvig’s General Ontology
Yet another view of general ontology structure is presented in Russell & Norvig’s book [38]. Every category of their ontology (see figure 2) is discussed in detail on example axioms. An example of this ontology in KIF [18] can be found at http://ltsc.ieee.org/suo/ ontologies/Russell-Norvig.txt.
10
Anything
AbstractObjects
Sets
Numbers
Categories
Events
RepresentationalObjects
Sentences
Measurements
Times
Intervals
Places
Moments
Things
Animals
Weights
PhysicalObjects Processes
Agents
Stuff
Solid
Liquid Gas
Humans
Figure 2: Russell & Norvig’s general ontology structure [38]
3.3
Ontology Engineering
Ontology engineering is a field in artificial intelligence or computer science that is concerned with ontology creation and usage. Report [31], that proposes and comments this field, declares that the ultimate purpose of ontology engineering should be “to provide a basis of building models of all things in which computer science is interested”. 3.3.1
Structure of Usage
An ontology can be divided into following subcategories according to [31] from the knowledge reuse and ontology engineering point of view as follows. This is rather a structure of ontologies from a point of view of their usage than a division of one general ontology. Some examples are included. • Workplace Ontology This is an ontology for workplace which affects task characteristics by specifying several boundary conditions which characterize and justify problem solving behaviour in the workplace. Workplace and task ontologies collectively specify the context in which domain knowledge is intended and used during the problem solving. Examples from circuit troubleshooting: fidelity, efficiency, precision, high reliability. • Task Ontology Task ontology is a system of vocabulary for describing problem solving structure of all the existing tasks domain independently. It does not cover the control structure. It covers components or primitives of unit inferences taking place during performing tasks. Task knowledge in turn specifies domain knowledge by giving roles to each objects and relations between them. Examples from scheduling tasks: schedule recipient, schedule resource, goal, constraint, availability, load, select, assign, classify, remove, relax, add. 11
• Domain ontology Domain ontology can be either task dependent or task independent. Task independent ontology usually relates to activities of objects. – Task-dependent ontology A task structure requires not all the domain knowledge but some specific domain knowledge in a certain specific organization. This special type of domain knowledge can be called task-domain ontology because it depends on the task. Examples from job-shop scheduling: job, order, line, due date, machine availability, tardiness, load, cost. – Task-independent ontology ∗ Activity-related ontology
· Object ontology. This ontology covers the structure, behaviour and function of the object. Examples from circuit boards: component, connection, line, chip, pin, gate, bus, state, role. · Activity ontology. Examples from enterprise ontology: use, consume, produce, release, state, resource, commit, enable, complete, disable.
∗ Activity-independent ontology
· Field ontology. This ontology is related to theories and principles which govern the domain. It contains primitive concepts appearing in the theories and relations, formulas, and units constituting the theories and principles. · Units ontology. Examples: mole, kilogram, meter, ampere, radian. · Engineering mathematics ontology. Examples: linear algebra, physical quantity, physical dimension, unit of measure, scalar quantity, physical components.
• General or Common ontology Examples: things, events, time, space, causality or behaviour, function etc. 3.3.2
Ontology Engineering Subfields
We can also divide the ontology or ontologies from the point of view of ontology engineering as a field. The subjects which should be covered by ontology engineering are demonstrated in [31]. It includes basic issues in philosophy, knowledge representation, ontology design, standardization, EDI, reuse and sharing of knowledge, media integration, etc. which are the essential topics in the future knowledge engineering. Of course, they should be constantly refined through further development of ontology engineering. • Basic Subfield – Philosophy(Ontology, Meta-mathematics) Ontology which philosophers have discussed since Aristotle is discussed as well as logic and meta-mathematics. 12
– Scientific philosophy Investigation on Ontology from the physics point of views, e.g., time, space, process, causality, etc. is made. – Knowledge representation Basic issues on knowledge representation, especially on representation of ontological stuff, are discussed. • Subfield of Ontology Design – General(Common) ontology General ontologies such as time, space, process, causality, part/whole relation, etc. are designed. Both in-depth investigation on the meaning of every concept and relation and on formal representation of ontologies are discussed. – Domain ontologies Various ontologies in, say, Plant, Electricity, Enterprise, etc. are designed. • Subfield of Common Sense Knowledge – Parallel to general ontology design, common sense knowledge is investigated and collected and knowledge bases of common sense are built. • Subfield of Standardization – EDI (Electronic Data Interchange) and data element specification Standardization of primitive data elements which should be shared among people for enabling full automatic EDI. – Basic semantic repository Standardization of primitive semantic elements which should be shared among people for enabling knowledge sharing. – Conceptual schema modeling facility (CSMF) – Components for qualitative modeling Standardization of functional components such as pipe, valve, pump, boiler, register, battery, etc. for qualitative model building. • Subfield of Data or Knowledge Interchange – Translation of ontology Translation methodologies of one ontology into another are developed. – Database transformation Transformation of data in a data base into another of different conceptual schema. – Knowledge base transformation Transformation of a knowledge base into another built based on a different ontology. • Subfield of Knowledge Reuse – Task ontology Design of ontology for describing and modeling human ways of problem solving.
13
– T-domain ontology Task-dependent domain ontology is designed under some specific task context. – Methodology for knowledge reuse Development of methodologies for knowledge reuse using the above two ontologies. • Subfield of Knowledge Sharing – Communication protocol Development of communication protocols between agents which can behave cooperatively under a goal specified. – Cooperative task ontology Task ontology design for cooperative communication • Subfield of Media Integration – Media ontology Ontologies of the structural aspects of documents, images, movies, etc. are designed. – Common ontologies of content of the media Ontologies common to all media such as those of human behavior, story, etc. are designed. – Media integration Development of meaning representation language for media and media integration through understanding media representation are done. • Subfield of Ontology Design Methodology – Methodology – Support environment • Subfield of ontology evaluation – Evaluation of ontologies designed is made using the real world problems by forming a consortium.
4
Using Ontologies
From above, we can see that an ontology can describe an upper-part of the knowledge base. The distinction between an ontology and a knowledge base is that ontology provides the basic structure or armature around which a knowledge base could be built. An ontology provides a set of concepts and terms for describing some domain, while a knowledge base uses those terms to represent what is true about some real or hypothetical world. Thus, a medical ontology might contain definitions for terms such a “leukemia” or “terminal illness”, but it would not contain assertions that a particular patient has some disease, although a knowledge base might. We can use the terms provided by the domain ontology to assert specific propositions about a domain or a situation in a domain. For example, in the electronic-device domain, we can represent a fact about a specific circuit, such as “circuit 35 has transistor 22 as a 14
component, where circuit 35 is an instance of the concept circuit and the transistor 22 is an instance of the concept transistor” [8]. Another example of blocks on table by Genesereth and Nillson is deeply discussed in [24] and [23] including discussion of possible ontologies. Once we have the basis for representing propositions, we can also represent knowledge involving propositional attitudes, such as hypothesize, believe, expect, hope, desire, fear, etc. Propositional attitude terms take propositions as arguments. For example, for the electronicdevice domain, we can assert for example “the diagnostician hypothesizes or believes that part 2 is broken, or the designer expects or desires that the power plant has an output of 20 megawatts”[8]. Thus, an ontology can represent beliefs, goals, hypotheses, and predictions about a domain, in addition to simple facts. The ontology also plays a role in describing such as plans and activities, because these also require specification of world objects and relations. Propositional attitude terms are also part of a larger ontology of the world, useful especially in describing the activities and properties of the special class of objects in the world called “intensional entities” — for example, agents like humans who have mental states.
4.1
Top-Level Ontology
Ontologies range in abstraction, from very general terms that form the foundation for knowledge representation in all domains, to terms that are restricted to specific knowledge domains. For example, space, time, parts, and subparts are terms that apply to practically all domains; malfunction applies to engineering or biological domains; and hepatitis applies only to medicine. Even in cases where a task might seem to be quite domain-specific, knowledge representation might include an ontology that describes knowledge at higher levels of generality. For example, solving problems in the domain of turbines might require knowledge expressed using domain-general terms such as flows or causality. Such general-level descriptive terms are called the upper-ontology or top-level ontology [8]. There are open research issues about the correct ways to analyze knowledge at the upper level. Illustration of different upper parts of ontologies are show in figure 1. For example, many ontologies have thing or entity as their root class. Figure 1 illustrates that thing and entity start to diverge at the next level. Some of the differences arise because not all of these ontologies are intended to be general-purpose tools, or even explicitly to be ontologies. Although differences exist within ontologies, general agreement exists between ontologies on many issues [8]: • There are objects in the world. • Objects have properties or attributes that can take values. • Objects can exist in various relations with other objects. • Properties and relations can change over time. • There are events that occur at different time instances. • There are processes in which objects participate and that occur over time. • The world and its objects can be in different states. 15
• Events can cause events or states as effects. • Objects can have parts. The representational repertoire of objects, relations, states, events, and processes does not say anything about which classes of these entities exist. The modeler of the domains makes these commitments.
4.2
Knowledge Sharing and Reuse
To support the sharing and reuse of formally represented knowledge among AI systems, it is useful to define the common vocabulary in which shared knowledge or ontology is represented. There have been several attempts to create engineering framework for constructing ontologies to support knowledge sharing and knowledge base reuse. 4.2.1
KIF
Michael R. Genesereth and Richard E. Fikes describe KIF (Knowledge Interchange Format), an enabling technology that facilitates expressing domain factual knowledge using a formalism based on augmented predicate calculus [18]. Knowledge Interchange Format is a computeroriented language for interchange of knowledge among disparate programs. Some important properties according to [18] are • it has declarative semantics, i.e. the meaning of expressions in the representation can be understood without appeal to an interpreter for manipulating those expressions • it is logically comprehensive, i.e. it provides for the expression of arbitrary sentences in the first-order predicate calculus • it provides for the representation of knowledge about the representation of knowledge • it provides for the representation of nonmonotonic reasoning rules • it provides for the definition of objects, functions, and relations 4.2.2
Ontolingua
Thomas R. Gruber has proposed a language called Ontolingua to help construct portable ontologies. In [22] an ontology as definitions of classes, relations, functions, and other objects is proposed for a specification of a representational vocabulary for a shared domain of discourse. The paper describes a mechanism for defining ontologies that are portable over representation systems. Definitions written in a standard format for predicate calculus are translated by a system called Ontolingua into specialized representations, including frame-based systems as well as relational languages (see figure 3). This allows researchers to share and reuse ontologies, while retaining the computational benefits of specialized implementations. The architecture of Ontolingua is shown in figure 3. Ontolingua enables to translate from a common ontology into several others. The same ontology can be used for different purposes in different systems. Instances of common representation idioms are recognized and transformed into a simpler, equivalent form using the second-order vocabulary from the Frame
16
Ontolingua
"Off the shelf" Ontology
Ontolingua definitions Parsing and syntax checking KIF sentences Recognition of idioms
Frame Ontology
LOOM Translator
LOOM T-Box descriptions
Epikit Translator
Epikit axioms
Other Ontolingua Translator
System-specific ontology
Canonicalization
Canonical Form
Pure KIF generator
Other KIF Translators
data models Prolog rules ...
Figure 3: Translation architecture for Ontolingua — translation from a common ontology into several others using the canonical form.
Ontology. These and other transformations result in a canonical form, which is given to backend translators that generate system-specific output. A pure KIF output is also available to be given to other translators developed for KIF, such as KIF-to-Prolog. 4.2.3
Collaboration
Ontologies facilitate collaboration among computer systems, among people, and between computers and people. Given an ontology for a particular domain it is possible to formalize exchanged or stored messages and so enable easier cooperation. Without such an ontology the terms used may not have the same meaning for all parties and a confusion may occur which can lead to misunderstanding. Also, when having formal form, it is possible for computers to assist in e.g. searching of needed information. The ontology for these purposes can be developed by one expert. The ontology is then usually fixed for other users. This has an advantage that the ontology is more likely precise [35]. On the other hand this approach requires some amount of time of an expert and a knowledge engineer. It is also possible to develop the ontology collaboratively. With this approach, everyone can contribute to the ontology. This is usually supported by web-based
17
tools, such as Ontolingua Server [22]. The ontology is then evolving according to actual needs, but can lead to a “wide” and even inconsistent ontology. One of the areas where the common terms to efficiently communicate are needed, is enterprise modeling. The enterprise ontology can support integration with existing and new tools and get together different views. Such an ontology also facilitates communication and information reuse, which can be supported by computer tools. An example of carefully designed enterprise ontology is described in [43]. On applications it is shown that the enterprise ontology encourages to use the terms in a unified form, and so the results by one group can be more easily reused by other group. Shared explicit conceptualization saves much efforts whenever collaborators from different areas have to work together. The effect of conceptualization is much bigger when the collaborators work together over large distances. The discussions usually consist of some proposals that are then revised and accepted or refuted. The discussion is usually supported by sketches, diagrams and other documents including formal structures. In Tadzebao World Wide Design Lab [46] such a discussion is supported by a set of “notepads” that can contain documents including ontologies. This tool even facilitates reusing previous solutions by retrieving them from other clients — the similar solution is searched by a central case-based reasoning engine. 4.2.4
Particular Ontologies Reuse
There were some attempts to develop ontologies for particular cases that would enable to construct knowledge bases from them. One of the areas that is being explored is diagnostics area — for example a careful ontological analysis of fault process and category of faults is described in [27]. Such an ontology can be core of a knowledge base solving any related problem. Using such a common knowledge base we can develop a deeper model of the system, not only a few heuristic shallow facts. This might enable to infer deeper causes of a fault. Another advantage is that the system with a model can easily explain its inferences. The ontology can be comparatively easily reused for diagnostics of other similar systems. The knowledge engineer doesn’t have to start from the scratch, but can use already predefined concepts in ontology. Also, because of using the same terms in the same ontology, knowledge bases from different sources can be more easily compared and possibly merged.
4.3
Communication in Multi-Agent Systems
Knowledge sharing and exchange is particularly important in multi-agent systems. An agent is usually described as a persistent entity with some degree of independence or autonomy that carries out some set of operations depending on what he perceives. An agent usually contains some level of intelligence, so it has to have some knowledge about it’s goals or desires. In multi-agent systems, an agent usually cooperates with other agents, so it should have some social and communicative abilities. Each agent has to know something about a domain he is working in and also has to communicate with other agents. An agent is able to communicate only about things that can be expressed by some ontology. This ontology must be agreed and understood among the agent community (or at least among its part) in order to enable each agent to understand messages from other agents.
18
Agent
Agent Platform Agent Management System
Agent Communication Channel
Directory Facilitator
Agent Communication Channel
Internal Platform Message Transport
Figure 4: FIPA Agent Management Reference Model [15]
The ontology in multi-agent system can be explicit or implicit. It is explicit when it is specified in declarative form e.g. as a set of axioms and definitions. It is implicit, when the assumptions on the meaning of its vocabulary are implicitly embedded in agents, i.e. in software programs representing agents. The explicit form enables and requires communication about an ontology that can be modified when agents agree on that. The implicit form is fixed, so no communication about ontology is required, but the change is impossible without reprograming agents. It is obvious that in open systems, where agents designed by different programmers or organizations may enter into communication, the ontology must be explicit. In these environments, it is also necessary to have some standard mechanism to access and refer to explicitly defined ontologies. We will describe a recommendation for this published by FIPA. This recommendation is being widely accepted and is also implemented in some systems for constructing agents (see for example FIPA Open Source at http://www.nortelnetworks. com/products/announcements/fipa/index.html [37] or JADE at http://sharon.cselt. it/projects/jade/ [4]). 4.3.1
FIPA Agent Management Model
The Foundation for Intelligent Physical Agents (FIPA, http://www.fipa.org/) is a nonprofit association registered in Geneva. It is formed by companies and organizations sharing the effort to produce specifications of generic agent technologies. FIPA’s purpose is to make and publish internationally agreed specifications or recommendations for agent systems and also to promote agent-based applications, services and equipments. Basic structure of the multi-agent system compliant to FIPA [15] is shown in figure 4. An agent is a fundamental actor on an agent platform. Agent platform (AP) consists of the machine(s), operating system(s), agent support software, FIPA compliant agent management components (DF, AMS, ACC – see below), proprietary internal platform message transport and agents. The communication between agents on one proprietary platform can be carried out in different ways, however according to FIPA at least the agent management components must be able to communicate using FIPA compliant communication protocols and languages. Agent management components (see figure 4) consist of these agents [15]: • Agent management system (AMS) is an agent that exerts supervisory control over 19
access to and use of the agent platform. Only one AMS must exist in a single AP. The AMS maintains a directory of logical agent names and their associated transport addresses for an agent platform. The AMS offers “white pages” services to other agents. • Agent communication channel (ACC) is an agent that provides the default communication methods between agents on different AP’s. One agent platform contains exactly one ACC agent. • Directory facilitator (DF) is an agent that provides “yellow pages” service to other agents. Agents may register their services with the DF and query the DF to find out what services are offered by other agents. One platform contains one main DF and can contain several other DF’s that help the main DF. These agent must be able to use FIPA agent communication language (ACL) [14] that is similar to KQML (Knowledge Query and Manipulation Language [2]). An example message in this language can be: (inform :sender agent1 :receiver hpl-auction-server :content (price (bid good02) 150) :in-reply-to round-4 :reply-with bid04 :language sl :ontology hpl-auction ) As we can see, this language is content independent3 , so that it can be used in any system, but when communicating, an agent must specify how to perceive the content. This specification is done via ontology. As was already said, there can be several ontologies and agents must be able to access and to refer to them. 4.3.2
Ontology Service by FIPA
For ontology services there is proposed a dedicated ontology agent (OA) in FIPA agent platform. The role of such an agent is to provide some or all of the following services [16]: • discovery of public ontologies in order to access them • help in selecting a shared ontology for communication • maintain (e.g. register with the DF, upload, download, or modify) a set of public ontologies • translate expressions between different ontologies and/or different content languages 3
• respond to query for relationship between terms or between ontologies
Some content languages such as SL, CCL, KIF and RDF are described in FIPA specification. An overview can be found in [17].
20
• facilitate the identification of a shared ontology for communication between two agents It is not mandatory that an ontology agent must provide all of these services, but every OA must be able to participate in a communication about these tasks. Also, it is not mandatory that every agent platform must contain an ontology agent, but when an ontology agent is present, it must complain to FIPA specifications. Example scenarios [16] of using OA particular services are: • Querying the OA for definitions of terms An user interface agent A wants to receive pictures from picture-archiver agent B to show them to a user. It asks agent B for “citrus”. However the agent B discovers that it doesn’t have any picture with that description. So it asks the appropriate OA to obtain sub-species of “citrus” within the given ontology. OA answers B that “orange” and “lemon” are sub-species of “citrus”, so the agent B can send pictures with these descriptions to agent A and so satisfy his requirements. • Finding equivalent ontology An ontology designer declares the ontology “car-product” to the ontology agent OA2 in U.S. in English terms and translates the same ontology to French for the ontology agent OA1 in France. Agent A2 uses the ontology from OA2 and wants to communicate with agent A1 about cars in ontology maintained by OA2. Because agent A1 doesn’t know ontology of agent A2, it queries OA1 for ontology equivalent to that one used by A2 (and maintained by OA24 ). OA1 returns its French ontology about cars and so A1 can inform A2 that these two ontologies are equivalent and that OA1 can be used as a translator. After that, a dialogue between A1 and A2 can start. • Translations of terms An agent A1 wants to translate a given term from an ontology #1 into the corresponding term in an ontology #2 (for example the concept “the name of a part” can be called name in ontology #1 and nomenclature in ontology #2). A1 queries DF for an OA which supports the translation between these ontologies. DF returns the name of an OA that knows the format of these ontologies (e.g. XML) and has capabilities to make translations between them. A1 can then query this OA and request translation of a term from ontology #1 to ontology #2. 4.3.3
Ontologies Relationships
In an open environment agents may benefit from knowing the existence of some relationships between ontologies, for instance to decide if and how to communicate with other agents. In the agent community, the ontology agent has the most adequate role to know that. It can be then queried for the information about such relationships and it can use that for translation or for facilitating the selection of a shared ontology for agent communication. In FIPA specification [16] the following relations are proposed: • Extension – ontology O1 extends ontology O2 The ontology O1 extends or includes the ontology O2. Informally this means that all 4 There is an ontology naming scheme described in [16] that allows to identify these ontologies — for example the English car ontology of agent OA2 can be named as OA2@http://makers.ford.com/car-product
21
the symbols that are defined within the O2 are found in the O1 together with the restrictions, meanings and other axiomatic relations of these symbols from O2. • Identical – ontologies O1 and O2 are identical Vocabulary, axiomatization and the language are physically identical, but the name can be different. • Equivalent – ontologies O1 and O2 are equivalent Logical vocabulary and logical axiomatization are the same, but the language is different (e.g. XML and Ontolingua). When O1 and O2 are equivalent then they are strongly translatable in both ways. • Strongly-translatable – source ontology O1 is strongly translatable to the target ontology O2 The vocabulary of O1 can be totally translated to the vocabulary of O2, axiomatization from O1 holds in O2, there is no loss of information from O1 to O2 and there is no introduction of inconsistency. Note that the representation languages can still be different. • Weakly-translatable – source ontology O1 is weakly translatable to the target ontology O2 The translation permits some loss of information (e.g. the terms are simplified in O2), but doesn’t permit introduction of inconsistency. • Approx-translatable – source ontology O1 is approximately translatable to the target ontology O2 The translation permits even introduction of inconsistencies, i.e. some of the relations become no more valid and some constraints do not apply anymore. The problem of deciding whether two logical theories (as ontologies usually are) have relationships to each other, is in general computationally very difficult. Therefore, knowing about these relationships often requires manual intervention and so a FIPA ontology agent should be also able to at least maintain database of these relationships. 4.3.4
FIPA Knowledge Model and Meta-Ontology
To allow agents to talk about knowledge and about ontologies, for instance to query for the definition of a concept or to define a new concept, a standard meta-ontology and knowledge model is necessary. This meta-ontology and knowledge model must be able to describe the primitives like concepts, attributes or relations. FIPA adopts for these purposes the OKBC Knowledge Model [16]. OKBC, the Open Knowledge Base Connectivity [9], provides operations for manipulating knowledge expressed in an implicit representation formalism called the OKBC Knowledge Model. This knowledge model supports an object-oriented representation of knowledge and provides a set of representational constructs and thus can serve as an interlingua for knowledge sharing and translation. The OKBC Knowledge Model includes constants, frames, slots, facets, classes, individuals, and knowledge bases. For precise description of the model, the KIF [18] is used. The OKBC knowledge model assumes a universe of discourse consisting of all entities about which knowledge is to be expressed. In every domain of discourse it is assumed that 22
all constants of the following basic types are always defined: integers, floating point numbers, strings, symbols, lists, classes. It is also assumed that the logical constants true and false are included in every domain of discourse. Classes are set of entities5 , and all sets of entities are considered to be classes. A frame is a primitive object that represents an entity in the domain of discourse. A frame is called class frame when it represents a class, and is called individual frame when it represents an individual. A frame has associated with it a set of slots that have associated a set of slot values. A slot has associated a set of facets that put some restrictions on slot values. Slots and slot values can be again any entities in the domain of discourse, including frames. A class is a set of entities, that are instances of that class (one entity can be instance of multiple classes). A class is a type for that entities. Entities that are not classes are referred to as individuals. Class frames may have associated a template slots and template facets that are considered to be used in instances of subclasses of that class. Default values can be also defined. Each slot or facet may contain multiple values. There are three collection types: set, bag (unordered, multiple occurrences permitted), and list (ordered bag). A knowledge base (KB) is a collection of classes, individuals, frames, slots, slot values, facets, facet values, frame-slot associations, and frame-slot-facet associations. KBs are considered to be entities of the universe of discourse and are represented by frames. There are defined standard classes, facets, and slots with specified names and semantics expressing frequently used entities. These are described in detail in [9] or [16]. Knowledge bases or ontologies conforming to OKBC are often expressed in KIF [18] or RDF [28]. An example of OKBC compliant editor of knowledge bases or ontologies that supports both of these formats is Prot´eg´e-2000 [19]. FIPA specification [16] defines ontology FIPA-meta-ontology based on OKBC Knowledge Model to describe ontologies. This ontology must be used by an agent when it talks about ontologies. Ontology FIPA-Ontol-service-Ontology must be used when requesting services of an ontology agent. This ontology extends the basic FIPA-meta-ontology by symbols enabling manipulation with ontologies. These ontologies are described in detail in [16].
4.4
Natural Language Understanding
One of AI fields that depend on a rich body of knowledge is natural language understanding (NLU) or natural language processing (NLP). Ontologies are useful in NLU in two ways. First, domain knowledge often plays a crucial role in disambiguation. A well designed domain ontology provides the basis for domain knowledge representation. In addition, ontology of a domain helps identify the semantic categories that are involved in understanding discourse of the domain. For this use, the ontology plays the role of a concept dictionary. In general, for NLU we need both a general-purpose upper ontology and a domain-specific ontology that focuses on the domain of discourse (such as military communications or business stories). CYC [45], Wordnet [3] and Sensus are examples of sharable ontologies that have been used for language understanding. NLU is one area that is highly motivating the work on ontologies — even CYC, which was originally motivated by the need of for the knowledge systems to have common world knowledge, has been tested more in natural language areas than in knowledge systems applications [8]. 5
The term class is used synonymously with the term concept as used in the description logic.
23
4.4.1
CYC NLP
The system for natural language developed with CYC in Cycorp is unique in having access to a very large, declaratively represented common sense knowledge base. CYC helps the natural language system handle word or phrase disambiguation, and also provides a target internal representation language (CycL [10]) that can be used to do interesting things, such as inference. A substantial portion of the CYC natural language processing system (the lexicon and many semantic rules) is actually represented in the CYC knowledge base. Syntactic parsing is carried out by application of phrase-structure rules to an input string. Semantic rules are applied to the output of the syntax module. It is in the application of the semantic rules that the knowledge in the knowledge base is proving especially advantageous. Most of the CYC pilot applications developed in the recent past have some NL component in their interfaces. The captioned image retrieval application [45], for example, accepts queries in English, and allows captioners to describe new images to the system using English sentences. The CYC NL team is currently expanding the lexicon, extending the parser, and adding new semantic capabilities to the system. 4.4.2
WordNet Database
Probably the most used linguistic database (and linguistic ontology) for natural language processing is WordNet [3] developed at Princeton University. WordNet is a lexical reference system whose design is inspired by current psycholinguistic theories of human lexical memory. English nouns, verbs, adjectives and adverbs are organized into synonym sets, each representing one underlying lexical concept. The WordNet includes several environments for manipulating its database. Each synonym set (called synset) is usually expressed as a unique combination of synonymous words. In general, each word can be associated to more than one synset and more than one lexical category. Different relations link the synonym sets. Examples of these are hypernymy, hyponymy, and antonymy. The first can be roughly assimilated to the usual subsumption relation, while the last links together opposite or inverse terms, such as tall/short. The WordNet offers two distinct services: a vocabulary, which describes the various word senses, and an ontology, which describes the semantic relationships among senses. Both these services can be used in natural language systems, such as WWW based systems (see below). Currently, an effort is being made for the translation of WordNet to several European languages in EuroWordNet project (see http://www.hum.uva.nl/~ewn/). For some languages, such as Dutch, Italian, Spanish, German, French, Czech and Estonian, the WordNet database is already available.
4.5
Document Search and Ontologies
To find a document in a large database is often not easy. For example, World Wide Web is so big knowledge base, that it is sometimes very time consuming to find some needed information. The search is usually entered in a form of a few keywords that are then searched. This is however not sufficient for today’s web — some king of more content-based search is needed. Structured content representations coupled with linguistic ontologies can increase both recall and precision of the search. Ontologies can be used in many ways for these purposes. They can support searching or knowledge mining from textual web sources, they can be used to specify the content of the 24
pages or to standardize content and query vocabulary. In this section we will also briefly describe applications for standardization of materials that need not be necessarily shared over the WWW, such as structured documents for business-to-business (B2B) information exchange. 4.5.1
OntoSeek
OntoSeek [25] is a tool intended for searching documents in product catalogs, such as yellow pages. Structured content representation with linguistic ontologies are used to increase both recall and precision of the retrieval. OntoSeek adopts a language of a limited expressiveness for content representation. It uses a large ontology based on WordNet (see above) for content matching. According to [25], we can distinguish three areas of information retrieval : • Text retrieval — the goal is to find relevant documents in a large collection in response to a user’s query expressed as a sequence of words. The user does not know much about collection content, therefore a precise semantic match between the query and the relevant documents is not very important. Also, it is assumed, that the textual quality of the documents is good. • Data retrieval — both the queries and data to be retrieved are encoded by a structured list of words, acting as values of a set of attributes established by the system’s designer. These words usually belong to a fixed taxonomy. However, large taxonomies can be hard to design and difficult to maintain. • Knowledge retrieval — here the query and data-encoding language is much more expressive. This results in increased precision, because the user can represent accurately the data’s content structure and formulate sophisticated queries. An arbitrary description can be then formed and matched on the basis of an ontology of primitive concepts and relations. However, this forces users to adopt a language that could be too expressive for their purposes. The last area is particularly interesting. In knowledge retrieval systems, an ontology provides the primitives needed to formulate queries and resource descriptions. Simple ontologies, such as keyword hierarchies, might be also useful to text and data retrieval techniques. OntoSeek uses WordNet and Sensus ontology for matching of queries and content. For content description (derived from natural language and refined by user) OntoSeek uses lexical conceptual graphs (LCG), that are simplified variants of Sowa’s conceptual graphs in way that they permit only “lexical” natural language relations between concepts (no ad-hoc relations such as part-of are allowed). The way of work with OntoSeek is following. In encoding phase, a resource description is converted into a LCG with the help of the user interface. The resulted LCG is stored into database. When querying, query description is converted into LCG with variables at places where the user expects an URL as an answer. Then the database is searched for matching graphs and the results are returned. Note that if the query language changes, but the underlying ontology for constructing LCGs remains the same, the system will continue to work correctly, i.e. if we change WordNet database to EuroWordNet database for some language, we have made the localization. 25
4.5.2
WebKB
Similar techniques for content matching are used in the project WebKB [30] (available at http://meganesia.int.gu.edu.au/~phmartin/WebKB/). Here however the content description is embedded directly to a web page. The description of a (part of) text is embedded into two HTML markers and . Because it is possible to use several formal languages to describe the content, the representational language must be also specified (e.g. for conceptual graphs). It is also possible to use alternative ways, such as embedding the description in the ALT property of images to describe content of images. The formal description language in WebKB include KIF [18], conceptual graphs, and the Resource Description Format (RDF) [28]. All of these suppose that some ontology is imported to distinguish between different concept, however it is possible to suppress the check whether the used concepts are already defined. Simpler notations, that are not so expressive, are also supported. These include structured text (with delimiters like “:”,“=>”,“[Person]->(Chrc)->[Age:*a]; }]; The WebKB tool provides a way to interpret these descriptions. It also provides an interface to Unix-like text processing commands to exploit web-accessible documents or databases and process them, for example query them. 4.5.3
Knowledge Representation Techniques
The current HTML standard in which the WWW pages are written doesn’t provide a way to embed some machine-understandable description of the content to the page. HTML was intended to provide a way to format documents to be readable for humans. There are however attempts to overcome this either by extending the standard or by using special existing standard markers for special purposes. One of the oldest ways of using HTML-based semantic markup to describe briefly the content of the page is to use HTML tags. In this way, we can embed the description of the whole page, that can be then processed by indexation or search agents. However this does’t provide any easily machine usable description. It is also usually used only for description of the whole page. There is a way to describe parts of the text using these tags and SPAN tags as shown in [44], but this is not standardized way. Another unstandardized proposed way is 26
to use Cascading Style Sheets (CSS), that could be used not only for formatting, but also for semantic markup. A standard enabling to embed any descriptive tree structure into a document is Extensible Markup Language XML [5]. XML is in fact meta-markup language, since it enables to define customs tags with desired syntactic structure via its DTD (Document Type Definition). The DTD can be viewed as an ontology specification. There are many proposals of DTD for specific areas — examples are CML, Chemical Markup Language and OML, Ontology Markup Language. The lexical structure of a XML document is a tree, but using shared identifiers it is possible to encode an arbitrary tree. W3C supported recommendation for semantic markup is Resource Description Framework (RDF) [28]. The data model of of RDF provides three entity types: • subjects is an entity that can be referred to by an address at the WWW. Subjects are the elements that are described by RDF objects. • predicate defines a binary relation between subjects and/or atomic values provided by primitive data type definitions in XML. • object specifies a value of a subject property. That is, objects provide the actual characterization of the WWW documents. There are alternative approaches using XML, such as SHOE or Ontobroker. SHOE (http://www.cs.umd.edu/projects/plus/SHOE/), a derivative of OML, provides tags for constructing ontologies and tags for annotating web documents. It doesn’t provide any standard top-level ontology, but it is possible to choose from offered particular ontologies. Ontobroker (http://ontobroker.aifb.uni-karlsruhe.de/) uses ontology based on frame logic to describe knowledge and to express queries. It includes an inference engine that can derive additional knowledge through inference rules. These two approaches have in common that they enable to express an “inferential knowledge”, i.e. to express relationships between entities in ontologies. These relationships can be used to form sophisticated queries or to infer new knowledge from existing web pages. Recently started DARPA program called DAML (DARPA Agent Mark-Up Language) [1] has a goal to create a mark-up language built upon XML that would allow users to provide machine-readable semantic annotations for specific communities of interest together with agent-based tools using this language. The program is using results and proposals mentioned above. So far the DAML results consists of the DAML language first proposal. The DAML language is a frame-based language based on RDF [28] and includes a basic top ontology for describing basic entities and ontologies. 4.5.4
Document Enrichment
So far we were concerned mostly with techniques enabling to represent document contents in a machine readable form to enable to use it for search. Ontologies can however be used for expressing not only the contents of documents, but also relations between documents. These relations can explicitly express what is usually expressed only implicitly in the contents of the document. For example in scholarly publications [40] a publication may support or refuse ideas expressed in another publication. Such a relationship with other documents adds a new information to the document and so this process is called “document enrichment” [41, 35]. 27
This additional information can be used for discovering relevant documents or for example for finding a potentially interesting gaps in available documents, such as finding possibly interesting themes for articles in a news system [12]. The particular relations between documents have to conform to an explicit structure to enable adding new documents to a structure and to enable using the structure. This explicit structure is expressed via ontology for a particular document domain. All the process of working with the documents is then ontology-driven [35]. The steps of a methodology described in [35] are: 1. Identify use scenario 2. Characterize viewpoint for ontology 3. Develop the ontology 4. Perform ontology-driven model construction 5. Customize query interface for semantic knowledge retrieval 6. Develop additional reasoning services on top of knowledge model The first three steps are performed by system authoring experts, because the quality of the document relationship model depends crucially on a quality of the ontology, that says what everything is possible to express. According to [35], the authors should focus on usability rather than on reusability. Reusable ontology often includes all potential aspects that could ever arise, however considering all these aspects when submitting a new document to a document structure could easily discourage system users. Usable ontology on the other side includes only highly relevant aspects that are easy to understand. The ontology is expressed in OCML (Operational knowledge modeling language) [34] language and used for rest of the steps above. For constructing ontologies a special web-based tool WebOnto can be used. The fourth step, construction of the model based on the ontology (i.e. document structure), is performed by casual users in a distributed way. When submitting a new document an author uses an environment enabling to define the relations to other documents or other aspects in a modeled world. This form-based environment called Knote is dynamically generated from the actual ontology. The fifth step, querying the “database”, is performed using Lois, a formbased interface for knowledge retrieval, that is also created automatically once the key classes for a knowledge model have been specified. This methodology summarized in [35] was used in several domains, such as enriching news stories [12], supporting scholarly debate [40] and knowledge management of medical guidelines. News server Planet-Onto [12] uses an ontology of events and classes like people, organizations, stories, projects and technologies. In addition to enriching and searching stories there are designed (but not implemented) two intelligent agent. NewsHound would gather data about popular news item and thus could solicit potentially popular stories by identifying gaps in the knowledge base. NewsBoy would enable personalized service for finding new stories according to user profile. Digital library server SholOnto [40] enables to contextualize ideas in relation to the literature. The ontology consists of contribution elements and relationships that are further
28
divided to argumentation and non-argumentation links. The ontology is designed to to support scholars in making claims by asserting relationships between concepts. Other scholars may support, raise-issue-with or refute these claims. Document enrichment can also be used to support organizational learning. The project Enrich [41] tries to support discussions structured along documents and concepts for learning. It is based on the fact, that the learning is more efficient when the acquired knowledge is immediately applied. The documents and domain concepts can be incrementally enriched by users, which supports further group learning.
4.6
Educational Systems
Computer educational or tutoring systems allow users to take a lesson without any time, place, or teacher availability constraints. Traditional computer aided instruction tools provide just a set of a static pages with text, or in better cases some simulators usable for one specific purpose. The courses delivered by these systems can be adapted and personalized usually only by a teacher — editor of the course. Also, these tools have the teaching strategies (if any) encoded directly in the taught content (such as “when the user presses button A, go to screen 23”). It is obvious, that it is practically impossible to reuse knowledge encoded in such a system. It is necessary to have the content knowledge separated from the system as well as from teaching strategies and so on. Intelligent tutoring systems (ITS, also called knowledge based tutors) are computer-based instructional systems that have separate data bases, or knowledge bases, for instructional content (specifying what to teach), and for teaching strategies (specifying how to teach), and attempt to use inferences about a student’s mastery of topics to dynamically adapt instruction. The adaptation of teaching materials for a particular student is usually made by artificial intelligence techniques. ITS usually requires to model the student and from the perceived model plan the next actions that would be the best for a particular student. These systems were usually built from the scratch up to date, without reusing any knowledge or parts of the other tutoring systems. Today however there are attempts to overcome this by establishing a common methodology or frameworks to enable knowledge reuse and so speed up development of such a systems and make their development easier. To establish common ontologies seems as the best way in order to achieve this goal. Some other advantages are discussed below. 4.6.1
EON
One of the pioneering works is EON [36], a collection of tools for authoring content, instructional strategy, student model and interface design. These authoring tools enable to store all the tutoring materials in a separate and thus a better reusable form. When editing a network of topic for further storing of course content, it is necessary to define an ontology of the network for a particular course area. That ontology can then be reused for another course in the same area. Also, the content can be easily transferred to another course with the same ontology with all of its properties. Other parts of the tutoring system are treated in the same way — it is necessary to define some concepts and constraints under which one will work and then it is possible to use them to specify the particular behaviour.
29
4.6.2
ABITS
Another attempt to create a reusable framework is ABITS [6], an agent based intelligent tutoring system. The knowledge taught is organized and indexed according so called Learning Object Metadata [26], that specifies properties and constraints of objects that could be considered as some entities that could be taught. For further description, resource description format [28] is used. Using a common format with the same ontology would facilitate sharing or transferring the knowledge to another tutoring systems. The topic structure is modeled via conceptual graphs. User modeling consists of cognitive state and learning preferences. Both these are expressed as fuzzy numbers. ABITS architecture consists of three types of agents — evaluation agents, that take care about the cognitive state of the student, affective agents, that evaluate learning preferences, and pedagogical agents, that update the curriculum (current plan for teaching). All these agents communicate with database connected to their type and with agents of other types. 4.6.3
Other Proposals
There are currently proposals and standards for some aspects of intelligent tutoring systems. IEEE initiative to specify Learning Object Metadata [26] was already mentioned. A partial task ontology for intelligent educational system is proposed in [33]. Such an ontology enables to provide a vocabulary in terms of which existing systems can be compared and enables to accumulate research results. Another advantages are that the educational tasks can be formalized and that it is possible to create reusable parts of tutoring systems. Also, it is possible to standardize communication protocol among component agents of ITS. Various entities, methods and concepts are analyzed to create an ontology that could be used for any ITS. A collection of other proposals of ontologies and their usage for several aspect of the intelligent educational systems, such as student model and preferences, curriculum ontology, task ontology, and others, can be found in [32]. Other initiatives are proposing ontologies for a particular areas that could be used for teaching special knowledge, such as simulation of physical systems. These ontologies have much in common with previously mentioned special ontologies and even top-level ontologies — they are often derived from them.
5
Conclusion
From the presented survey it follows that ontologies will fundamentally change the way in which systems are constructed. Today, knowledge bases are still built with little sharing or reuse — almost each one starts from a blank slate. In the future, intelligent systems developers will have libraries of ontologies at their disposals. Rather than building from scratch, they will assemble knowledge bases from components drawn from the libraries. This should greatly decrease development time while improving the robustness and reliability of the resulting knowledge bases. There is proposed a new field within knowledge engineering that is called “ontology engineering” [31]. This field is concerned with all theoretical or practical aspects of ontologies, such ontology development, maintenance, reuse etc. We have discussed some of the areas of interest within this discipline. 30
An ontology can be viewed (among other possibilities) as a topmost part of the knowledge base. This upper part is usually encoding the general or “common sense” knowledge, such as that “objects can have properties”. For larger applications, that are not strictly focused on one simple thing, such a knowledge can be very extensive and difficult to maintain. One of the examples that were shown is the project CYC [45], that claims to encode substantial part of the people’s common sense knowledge. We have shown applications of ontologies in several areas. The first applications, that started broader interest in ontologies, were the ones in knowledge sharing and reuse [22]. Ontologies here help to express commonly accepted conceptualization of domains. The worlds in these domains can be described in various ways or languages, but if this all conform to one ontology, then it is much more easily possible to provide translation between different descriptions. Ontology helps here as interlingua, that is that it serves as a common language for translations (we do not have to construct translators for every pair of language, since the translators between each language and common interlingua are enough to provide any translation). This facilitates sharing of knowledge and also reuse of the knowledge in various systems, that do not have to use exactly the same internal representation of the world. An area that is tightly related to the knowledge sharing and reuse is communication in multi-agent systems. If agents have to communicate, they have to use a commonly agreed and understood syntax and semantic of the messages. If this is expressed explicitly as an ontology, it is possible to manipulate with various forms of communication. For example it is possible to find a translation between ontologies, to query meanings in other ontologies, to enrich ontology of an agent and so on. We have shown, how this is proposed to be done in FIPA [16] standards through a dedicated ontology agent. In the area of natural language processing the ontology is usually perceived as a structure of lexical word meanings. The ontology plays here the role of the conceptual dictionary. Probably the most known linguistic database, WordNet [3], was described. It should be mentioned that such ontologies are not only vocabularies of words. They in addition include semantic and other relationships among senses and meanings of each word. An application that needs to have an explicit conceptualization is a sophisticated search of the document in a large database, such as World Wide Web. If we want to search for documents not only by entering a few keywords, but also use a structure encoded in and between documents, we have to build a database of such structures together with the simple directory of the documents. Among the systems using the conceptual structures encoding the internal semantic structure of the documents we have mentioned OntoSeek [25] and WebKB [30]. Both these projects use ontologies for defining possible structure of the content description of the document. It is then possible to search by queries that use these structures. We have shown some attempts for standardization of such a description for the WWW, often built on XML [5]. The way of introducing relations between document is described as “document enrichment” [40, 35]. This concept of work was described on domains like scholarly debate, news stories, medical guidelines, and organizational learning. The last presented application that needs explicitly defined ontologies, is an area of intelligent educational systems. We have shortly described two frameworks for intelligent tutoring systems (or knowledge based tutors), EON [36] and ABITS [6]. On both these systems it is shown, that an ontology can help to increase speed of creating tutoring system for particular domain as well as reuse of already created parts of the systems. There are already arising committees for standards within these educational systems initiatives, similarly like in the document search area mentioned above. 31
To conclude, we can see, that the explicit conceptualization of a domain in a form of ontology facilitates knowledge sharing, knowledge reuse, communication and collaboration and construction of knowledge rich and intensive systems. There are some methodologies how to construct and maintain ontologies and how to use them in practical applications, but this is rather a set of recommendations then exact theories. Despite this, ontologies are being more and more widely used for construction of systems that require an explicitly encoded knowledge.
Acknowledgements This research has been supported by the MSMT Grant No. 212300013. I would like to thank professor Vladim´ır Maˇr´ık for his valuable comments to the early versions of this work.
References [1] DAML - DARPA Agent Mark-Up Language. http://www.daml.org/. [2] UMBC KQML Web. http://www.cs.umbc.edu/kqml/. [3] WordNet - a Lexical Database for English. http://www.cogsci.princeton.edu/~wn/. [4] Fabio Bellifemine, Agostino Poggi, and Giovanni Rimassa. JADE – A FIPA-Compliant Agent Framework. In Proceedings of PAAM’99, pages 97–108, London, 1999. [5] Tim Bray, Jean Paoli, and C. M. Sperberg-McQueen. Extensible Markup Language (XML) 1.0. http://www.w3.org/TR/REC-xml, 1998. [6] Nicola Capuano, Marco Marsella, and Saverion Salerno. ABITS: An Agent Based Intelligent Tutoring System for Distance Learning. In C. Peylo, editor, Proceedings of the International Workshop on Adaptive and Intelligent Web-Based Educational Systems Held in Conjunction with ITS 2000, Montreal, Canada, 2000. [7] B. Chandrasekaran, J. R. Josephson, and V. Richard Benjamins. The Ontology of Tasks and Methods. In Knowledge Acquisition, Modeling and Management, 1998. [8] B. Chandrasekaran, John. R Josephson, and Richard V. Benjamins. What Are Ontologies, and Why do We Need Them? IEEE Transactions on Intelligent Systems, pages 20–26, 1999. [9] Vinay K. Chaudhri, Adam Farquhar, Richard Fikes, Peter D. Karp, and James P. Rice. Open Knowledge Base Connectivity 2.0.3. http://www.ai.sri.com/ okbc/spec.html, 1998. [10] CYCORP. The CycL Language. http://www.cyc.com/cycl.html. [11] Randall Davis, Howard Shrobe, and Peter Szolovits. What is a Knowledge Representation? AI Magazine, 14:17–33, 1993. [12] John Domingue and Enrico Motta. A Knowledge-Based News Server Supporting Ontology-Driven Story Enrichment and Knowledge Retrieval. In 11th European Workshop on Knowledge Acquistion, Modelling, and Management (EKAW’99), 1999. 32
[13] Edward Feigenbaum, Robert Engelmore, Thomas Gruber, and Yurni Iwasaki. Large Knowledge Bases for Engineering: The How Things Work Project of the Stanford Knowledge Systems Laboratory. Technical Report KSL 90-83, Standford University, Knowledge System Laboratory, 1990. [14] FIPA. Agent Communication Language Specification. http://www.fipa.org/specs/ fipa00003/, 1997. [15] FIPA. Agent Management Specification. http://www.fipa.org/specs/fipa00002/, 1998. [16] FIPA. Ontology Service Specification. http://www.fipa.org/specs/fipa00006/, 1998. [17] FIPA. Content Languages Specification. http://www.fipa.org/specs/fipa00007/, 2000. [18] Michael R. Genesereth and Richard E. Fikes. Knowledge Interchange Format, Version 3.0, Reference Manual. http://logic.stanford.edu/kif/Hypertext/kif-manual. html, 1994. [19] W. E. Grosso, H. Eriksson, R. W. Fergerson, J. H. Gennari, S. W. Tu, and M. A. Musen. Knowledge Modeling at the Millennium — The Design and Evolution of Protege2000. Technical Report SMI-1999-0801, Stanford Medical Informatics, 1999. http: //smi-web.stanford.edu/pubs/SMI_Abstracts/SMI-1999-0801.html. [20] Thomas R. Gruber. What is an Ontology? what-is-an-ontology.html.
http://www-ksl.stanford.edu/kst/
[21] Thomas R. Gruber. A Translation Approach to Portable Ontology Specifications. Knowledge Acquisition, 1993. [22] Thomas R. Gruber. Toward Principles for the Design of Ontologies Used for Knowledge Sharing. In Nicola Guarino and Roberto Poli, editors, Formal Ontology in Conceptual Analysis and Knowledge Representation. Kluwer Academic Publishers, 1993. [23] Nicola Guarino. Understanding, Building and Using Ontologies. International Journal of Human-Computer Studies, special issue on Using Explicit Ontologies in KBS Development, 1997. [24] Nicola Guarino and Pierdaniele Giaretta. Ontologies and Knowledge Bases - Towards a Terminological Clarification. In N.J.I. Mars, editor, Towards Very Large Knowledge Bases. IOS Press, Amsterdam, 1995. [25] Nicola Guarino, Claudio Masolo, and Guido Vetere. OntoSeek: Content-Based Access to the Web. IEEE Transactions on Intelligent Systems, pages 70–80, 1999. [26] Wayne Hodgins. Learning Objects Metadata, Working Draft. Technical report, IEEE Learning Technology Standartization Committee, 2000. http://ltsc.ieee.org/doc/ wg12/LOM_WD4.htm. [27] Yoshinobu Kitamura and Riichiro Mizoguchi. An Ontological Analysis of Fault Process and Category of Faults. In Proc. Of Tenth International Workshop on Principles of Diagnosis, pages 118–128, 1999. 33
[28] Ora Lassila and Ralph R. Swick. Resource Description Framework (RDF) Model and Syntax Specification. http://www.w3.org/TR/PR-rdf-syntax/, 1999. [29] Douglas Lenat. The Dimensions of Context-Space. CYCORP, Austin, Texas, 1998. [30] Philippe Martin and Peter W. Eklund. Knowledge Retrieval and the World Wide Web. IEEE Transactions on Intelligent Systems, pages 18–25, 2000. [31] Riichiro Mizoguchi and Mitsuru Ikeda. Towards Ontology Engineering. Technical Report AI-TR-96-1, The Institute of Scientific and Industrial Research, Osaka University, 1996. [32] Riichiro Mizoguchi and Tom Murray, editors. AI-ED 99 Workshop on Ontologies for Intelligent Educational Systems, 1999. http://www.ei.sanken.osaka-u.ac.jp/aied99/ aied99-onto.html. [33] Riichiro Mizoguchi, K. Sinitsa, and Mitsuru Ikeda. Knowledge Engineering of Educational Systems for Authoring System Design - A preliminary results of task ontology design. In Proceedings of EAIED, pages 329–335, Lisabon, 1996. [34] Enrico Motta. An Overview of the OCML Modelling Language. In Proceedings of the 8th Workshop on Knowledge Engineering Methods and Languages (KEML’98), 1998. [35] Enrico Motta, Simon Buckingham Shum, and John Domingue. Ontology-Driven Document Enrichment: Principles and Case Studies. International Journal of HumanComputer Studies, 52:1071–1109, 2000. [36] Tom Murray. Authoring Knowledge Based Tutors: Tools for Content, Instruction Strategy, Student Model, and Interface Design. Journal of the Learning Sciences, 7(1):5–64, 1998. [37] Nortel Networks. FIPA-OS V1.3.3 Distribution Notes, 2000. [38] Stuart Russell and Peter Norvig. Artificial Intelligence: A Modern Approach. PrenticeHall International, Inc., 1995. ISBN 0-13-360124-2. [39] James Schoening. IEEE Standard Upper Ontology Study Group Homepage. http: //ltsc.ieee.org/suo/, 2000. [40] Simon Buckingham Shum, Enrico Motta, and John Domingue. ScholOnto: An OntologyBased Digital Library Server for Research Documents and Discourse. International Journal on Digital Libraries, 3(3), August/Sept. 2000. [41] Tamara Sumner, John Domingue, Zdenek Zdrahal, Marek Hatala, Alan Millican, Jayne Murray, Knut Hinkelmann, Ansgar Bernardi, Stefan Wess, and Ralph Traphoner. Enriching Representations of Work to Support Organisational Learning. In Proceedings of the Interdisciplinary Workshop on Building, Maintaining, an Using Organizational Memories (OM-98), 13th European Conference on Artificial Intelligence (ECAI-98), Bringhton, UK, 1998. [42] William Swartout. Ontologies. IEEE Transactions on Intelligent Systems, pages 18–19, 1999.
34
[43] Mike Uschold, Martin King, Stuart Moralee, and Yannis Zorgios. The Enterprise Ontology. Technical Report AIAITR -195, Artificial Intelligence Applications Institute, University of Edinburgh, 1998. [44] Frank Van Harmelen and Dieter Fensel. Practical Knowledge Representation for the Web. In IJCAI99 Workshop on Intelligent Information Integration, 1999. [45] David Whitten. The Unofficial, Unauthorized Cyc Frequently Asked Questions. http://www.robotwisdom.com/ai/cycfaq.html, 1997. [46] Zdenek Zdrahal and John Domingue. The World Wide Design Lab: An Environment for Distributed Collaborative Design. In Proceedings of International Conference on Engineering Design, Tampere, 1997.
35
