HISTORY OF ONTOLOGICAL TECHNOLOGY AND A MODERN MEDICAL APPLICATION

HISTORY OF ONTOLOGICAL TECHNOLOGY AND A MODERN MEDICAL APPLICATION

IOANNA MALAGARDI

Medical Informatics Laboratory, School of Nursing, National and Kapodistrian University of Athens, Athens, Hellas

Journal of Neural, Parallel and Scientific Computations. Vol 14 2005, pp.491-505.

ABSTRACT: A brief history of the new field of study called “Ontological Engineering” or “Ontological Technology” is presented in the present paper. A modern bilingual ontology-based system for Medical Informatics is also presented in the present paper. The roots of the idea of an Ontology in Ancient Greek Philosophy and particularly in the works of Aristotle are traced. Ontology as a branch of philosophy is considered as the science of what is, of the kinds and structures of objects, properties, events, processes and relations in every area of reality. Our system although based on these ancient primitives it uses modern Natural Language Processing (NLP) methods such as subject ellipsis resolution, implicit semantic relation determination and analysis of complex terms into their components. An important novelty of our system is the generation of explanations that support medical terminology management. Our system consists of a subsystem with a friendly natural language user interface, a subsystem for the resolution of the ellipsis of the agent of an action, a subsystem for the discovery of implicit semantic relations between the elements of a noun phrase, a subsystem of compound medical terms and a subsystem for hypernym extraction from term definitions.

Keywords- Medical Informatics, Ontology, Natural Language Processing, Explanation

1. INTRODUCTION

A brief history of the new field of study called “Ontological Engineering” or “Ontological Technology” is presented in the present paper. A modern bilingual ontology-based system for Medical Informatics with a friendly natural language interface is also presented in the present paper. The system applies to medical informatics Natural Language Processing (NLP) methods such as subject ellipsis resolution, implicit semantic relation determination and analysis of complex terms into their components. An important novelty of the system is the generation of explanations that support medical terminology management. The technology of explanation generation was developed from our work on intelligent text mining and knowledge discovery from biomedical texts (Kontos et al., 1998, 1999, 2000a, 2000b, 2000c, 2004). The recent introduction of new computer-based medical systems has made the problem of this technology even more complex. One reason for this added complexity is the introduction of many genomics, proteomics and engineering terms that are often combined with traditional biological and medical terms. These new complex and hybrid terms create some severe difficulties in understanding and learning to use computer tools to manage medical ontology with NLP software. Some examples of work of medical ontology applications (Rossi Mori et al., 1997; Schulze-Kremer, 1997; Schubert, 2001) are related to the present work. Our system consists of the following subsystems:

A subsystem with a friendly natural language user interface.

A subsystem for the resolution of the ellipsis of the agent of an action.

A subsystem for the discovery of implicit semantic relations between the elements of a noun phrase.

A subsystem of compound medical terms.

A subsystem for hypernym extraction from term definitions.

2. A BRIEF HISTORY OF ONTOLOGICAL TECHNOLOGY

Ontological Technology is a new branch of Information Technology that develops methods and tools for the construction and processing of computer based ontologies. These computer based ontologies are used for the support of various systems such as question answering and text analysis for information extraction and knowledge mining by computer.

The idea of an Ontology has its roots in Ancient Greek Philosophy and particularly in the works of Aristotle. Ontology as a branch of philosophy is the science of what is of the kinds and structures of objects, properties, events, processes and relations in every area of reality. “Ontology” is often used by philosophers as a synonym of “metaphysics” (literally: “what comes after the Physics”), a term which was itself used by early students of Aristotle to refer to what Aristotle himself called “first philosophy”.
The term “ontology” (or ontologia) was coined in 1613, independently, by two philosophers, Rudolf Göckel (Goclenius), in his Lexicon philosophicum and Jacob Lorhard (Lorhardus), in his Theatrum philosophicum. Its first occurrence in English as recorded by the Oxford English Dictionary appears in Bailey’s dictionary of 1721, which defines ontology as “an Account of being in the Abstract”.

In a development outside philosophy an activity has been advanced in recent years in certain extra-philosophical disciplines, as linguists, psychologists and anthropologists have sought to elicit the ontological commitments (“ontologies”, in the plural) of different cultures and groups. Researchers in psychology and anthropology have sought to establish what individual human subjects, or entire human cultures, are committed to, ontologically, in their everyday cognition in much the same way in which philosophers of science had attempted to elicit the ontological commitments of the natural sciences. Alternatively, they have engaged in inquiries designed to establish how folk ontologies develop through infancy and childhood, or to establish the degree to which given elements of folk ontologies reflect universal features of the human cognitive system.

Ontology seeks to provide a definitive and exhaustive classification of entities in all spheres of being. The classification should be definitive in the sense that it can serve as an answer to such questions as: What classes of entities are needed for a complete description and explanation of all the activities in the universe? Or: What classes of entities are needed to give an account of what makes true all truths? Or: What classes of entities are needed to facilitate the making of predictions about the development of the behaviour of a physical or biological model? (Smith, 2001).

Ontological Technology aims at answering such questions by producing computer implementation methods that lead to the construction of information systems that utilize computer based ontologies for the accomplishment of their task.

Some of the early implementations of computer based ontologies that prepared the appearance of Ontological Technology were reported as parts of Question Answering Systems (Kontos, 1972). In these pioneering systems rules were implemented that stated relations between semantic categories. These semantic categories (Kontos, 1972) corresponded to the basic elements required to formally state facts and describe processes in the actual or some possible world. Such elements were the notions of objects, functions, sets, properties and relations. It was then required that the programming language used for the implementation of Question Answering Systems is capable of expressing everything that can be stated using the semantic categories of a given universe of discourse, possible world or problem domain.

One early influential use of the term “ontology” in the computer science community was by the (McCarthy, 1980) theoretical paper on “circumscription”. McCarthy argues in this paper that the proper treatment of common-sense reasoning requires that common-sense knowledge be expressed in a form which will allow us to express propositions like “a boat can be used to cross rivers unless there is something that prevents its use”. This means, he says, that:

“We must introduce into our ontology (the things that exist) a category that includes something wrong with a boat or a category that includes something that may prevent its use… Some philosophers and scientists may be reluctant to introduce such things, but since ordinary language allows “something wrong with the boat” we shouldn’t be hasty in excluding it”.

Another early use of the term is in the writings of McCarthy’s student Patrick Hayes, “Naive Physics Manifesto” (Hayes, 1979), in which he describes a new paradigm for Artificial Intelligence research, which represents a move away from the procedural modelling of reasoning processes and towards the construction of systems embodying large amounts of declarative knowledge. It should be noted, however, that Hayes conceives his project as that of formalizing not the physical features themselves but rather our “mental models” – so that his “Naive Physics Manifesto” is a contribution not to the discipline of ontology in the traditional sense at all, but rather to that of knowledge representation. In 1985 Hayes published his “Second Naïve Physics Manifesto” and his “Ontology for Liquids”, revised versions of the earlier work (Hayes, 1985a, 1985b).

In his second manifesto Hayes talks not of “faithfulness to reality” but rather of the ability “to interpret our axioms in a possible world”. To establish whether these axioms are true or not means to develop “an idea of a model of the formal language in which the theory is written: a systematic notion of what a possible world is and how the tokens of the theory can be mapped into entities … in such worlds”. The idea that an ontology has to do with what entities are included in a model, or in a possible world, is present also in the writings of J. Sowa (1984), who refers to “an ontology for a possible world – a catalogue of everything that makes up that world, how it’s put together, and how it works”.

Already by 1986 (Alexander et al., 1986), the usage of “ontology” as meaning just “conceptual model” had been established in knowledge representation research. In (Gruber, 1993) this new sense of the term is assumed in his definition of an ontology as “the specification of a conceptualization”. He is referring in this phrase to a technical notion of “conceptualization” introduced by (Genesereth & Nilsson, 1987), where conceptualizations are conceived as extensional entities defined in terms of sets of relations of certain sorts. For present purposes, however, we can gain a sufficiently precise understanding of the nature of ‘ontology’ as Gruber conceives it if we rely simply on the account he himself gives in passages such as the following:

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 (Gruber, 1995).

In a scientific setting we might use concepts such as virus and nitrous oxide. Such conceptualizations are often tacit; that is, they are often not exposed in any systematic way. But tools can be developed to specify and to clarify the concepts involved and to establish their logical structure, and thus to render explicit the underlying taxonomy. It is for this reason that the ontological engineer aims not for truth, but rather, merely, for adequacy to whatever is the pertinent domain as defined by the texts. The goal is not truth relative to some independently existing domain of reality, which is after all often hard to achieve, but merely truth relative to some conceptualisation. Some recent developments are presented in (Guarino, 1995, 1999) and (Guarino & Welty, 2000).

3. SYSTEM ARCHITECTURE

The architecture of a modern system that we have implemented consists of the following subsystems that are described below:

A subsystem with a friendly natural language user interface.

A subsystem for the resolution of the ellipsis of the agent of an action.

A subsystem for the discovery of implicit semantic relations between the elements of a noun phrase.

A subsystem of compound medical terms.

A subsystem for hypernym extraction from term definitions.

The first subsystem is a friendly natural language interface developed in order to coordinate the above subsystems into an integrated system that supports the medical terminology management system.

The second subsystem is based on a method of choosing the subject of a verb that denotes the agent of an action between more than one candidate by using ontological and other domain knowledge that is introduced beforehand into the subsystem.

The third subsystem is based on a method developed for the discovery of implicit semantic relations between elements of a noun phrase supported by a domain specific ontology. Possible implicit semantic relations are given to the system at the highest possible level of the ontology between the general concepts at that level. The elements of each noun phrase are categorized using the ontology and using appropriate inheritance rules the higher-level relations are assigned to the lower level concepts.

The fourth subsystem is based on a method of composing and decomposing compound medical terms particularly of Greek origin, as well as their definitions.

The fifth subsystem is based on a method of hypernym extraction from term definitions.

3.1. A subsystem with a friendly natural language user interface

The first subsystem is a friendly natural language interface developed in order to coordinate the above subsystems into an integrated system that supports the use of machine-readable dictionaries and ontologies for the construction of medical terminology management systems. The main novelty of this subsystem is the generation of explanations that support NLP methods for medical terminology management. The system includes a user interface for the retrieval of groups of terms that have the same hypernym as well as for the insertion of new terms either in Greek or in English. This interface is mainly used for the management of our Greek-English medical lexicon.

This lexicon was created in the context of the system is to support information extraction from medical texts. The lexicon includes about 10.000 terms. The terms have been translated from English using a number of English to Greek special purpose dictionaries. These dictionaries have been created for uses from various disciplines such as medicine and biology. The translation to Greek differed in many cases between the different dictionaries used and the choice of the best translation had to be made in consultation with medical doctors we collaborate with.

3.2. A subsystem for the resolution of the ellipsis of the agent of an action

The second subsystem is based on a method of choosing the subject of a verb that denotes the agent of an action between more than one candidate by using ontological and other domain knowledge that is introduced beforehand into the subsystem. The details of this subsystem will be reported in a future paper.

3.3. A subsystem for the discovery of implicit semantic relations between the elements of a noun phrase

The third subsystem is based on a method developed for the discovery of implicit semantic relations between elements of a noun phrase supported by a domain specific ontology (Malagardi, 1996). Possible implicit semantic relations are given to the system at the highest possible level of the ontology between the general concepts at that level. The elements of each noun phrase are categorized using the ontology and using appropriate inheritance rules the higher-level relations are assigned to the lower level concepts. Some examples of the computation of implicit relations of the parts of compound medical terms are given below together with the explanations generated.

[“Chondrosarcoma”]= [“sarcoma at chondros”]

explanation:

The compound consists of entity-nouns “sarcoma” which is a tumor and “chondros” which is a body part. Since tumors are located in body parts it follows that the implicit relation is that of location where chondros is the location of the sarcoma.

[“Hemarthrosis”]= [“hema with origin arthrosis”]

explanation:

The compound consists of entity-nouns “hema” which is a fluid and “arthrosis” which is a body part. Since liquids may have as origin body parts it follows that the implicit relation is that of origin where arthrosis is the origin of hema.

[“Staphylococcus”]= [“coccus appearing like staphylo”]

explanation:

The compound consists of entity-nouns “staphylo” which is a fruit and “coccus” which is a microorganism. Since fruits and microorganisms may have only common appearence it follows that the implicit relation is appearance where the coccus appears like staphylo.

[“Azotemia”]= [“azote inside ema”]

explanation:

The compound consists of entity-nouns “azot” which is a gas and “hema” which is a fluid. Since gases maybe contained in fluids it follows that the implicit relation is that of content where azot is inside hema.

[“Bacteremia”]= [“bacter inside ema”]

explanation:

The compound consists of entity-nouns “bacter” which is a microorganism and “hema” which is a fluid. Since microorganisms maybe contained in fluids it follows that the implicit relation is that of content where azot is inside hema.

In order to be able to perform the computation of implicit relations similar to the ones mentioned above our system must have access to a knowledge base of extra linguistic knowledge derived from the medical domain. Both taxonomic and meronomic extra linguistic knowledge is used as knowledge base that supports the analysis of the phrases that are being processed by the system. Such a knowledge base will also contain rules of the following kind for the computation of the implicit relations:

If tumors are located in body parts it follows that the implicit relation is that of location.

If liquids may have as origin body parts it follows that the implicit relation is that of origin.

If fruits and microorganisms may have only common appearence it follows that the implicit relation is an attributive relation of appearance.

If gases maybe contained in fluids it follows that the implicit relation is that of content.

microorganisms maybe contained in fluids it follows that the implicit relation is that of content.

If harmful substances damage body parts it follows that the implicit relation is that of causality.

If comparative relations may exist between body parts it follows that the implicit relation is that of comparison.

If body parts have size it follows that the implicit relation is an attributive relation.

If sensing can be rated it follows that the implicit relation is an attributive relation.

If gases can be moved it follows that the implicit relation is a relation between a process and a moving object.

If diseases are located in body parts it follows that the implicit relation is a locative relation between a disease and a body part.

If discolorations can apply to body parts it follows that the implicit relation is a patient relation between the discoloration and the body part.

If surgical procedures apply to body parts it follows that the implicit relation is a patient relation between the procedure and the body part.

If instruments are used for monitoring physiologic processes it follows that the implicit relation is that of use.

3.4. A subsystem of compound medical terms

The fourth subsystem described in the present section is based on a method of composing and decomposing compound medical terms particularly of Greek origin, as well as their definitions. The subsystem accomplishes the analysis of compound medical terms of Greek origin expressed either in Greek or English into their constituent words. The final aim of this analysis is the automatic synthesis of the definition of the compound word from the definitions of the constituent words.

The components of Greek origin compound words are found by the application of a computer program on each compound. This program uses a partial matching algorithm that uses a number of rules for recognizing the components of the compound term.

The definitions of all possible senses of each compound term may be generated using a lexicon defining in natural language the senses of their component words. The ontology implicit in the definitions is used for choosing the appropriate rule of definition combination. A number of rules specifying the relations allowable between general concepts that group the component words also support this generation. A Prolog program was implemented as part of the subsystem described here that is related to the one computing implicit relations between the components words in a sublanguage using linguistic and extra linguistic knowledge as in Malagardi (1996).

An extra linguistic knowledge base containing ontological knowledge derived from the domain or microcosm of the sublanguage is used in the program. This knowledge base may be used for the computation of implicit relations between nouns. The definitions of compound terms can be synthesized automatically. This synthesis is accomplished using the appropriate definitions and the appropriate parts of the ontology of their components. Some examples of definitions of component words are given below as coded in the program using the predicate “translates”:

translates([acro],[the,extremity_ties]).

translates([cyanosis],[a,bluish,discoloration]).

translates([aero],[air]).

translates([brady],[abnormal,slowness]).

translates([brady],[sluggineshness]).

translates([cardia],[the,heart]).

translates([kinesia],[movement]).

translates([kinesia],[the,physical,and,mental,response_es]).

translates([pnea],[the,breathing]).

translates([genic],[caused,by,abnormal,function]).

translates([megaly],[large,size]).

translates([pathy],[disease]).

translates([cardio],[the,heart]).

translates([myo],[the,muscle]).

Part of the ontology used for processing these definitions is coded in the program as follows:

isa([acro],entity).

isa([aero],entity).

isa([cardio],entity).

isa([cardia],entity).

isa([brady],property).

isa([cyanosis],process).

isa([phagy],process).

isa([kinesia],process).

isa([pnea],process).

isa([genic],process).

isa([megaly],property).

isa([myo],entity).

isa([pathy],process).

Some examples of definitions of compound terms synthesized automatically by the module described in the present section are given below together with an explanation of the process that is automatically produced. This synthesis is accomplished using the appropriate definitions and the appropriate parts of the ontology of their components.

["acro"]+["cyanosis"]=["a","bluish","discoloration","of","extremities"]

explanation:

The compound consists of one entity-noun i.e. “acro” which is a body part and one process-noun i.e. “cyanosis” which is a discoloration. Since we know that discolorations can apply to body parts it follows that the implicit relation is a patient relation between the discoloration and the body part.

["aero"]+["phagy"]=["swallowing","of","air"]

explanation:

The compound consists of one entity-noun i.e. “aero” which is a gas (in reality a mixture of gases) and one process-noun i.e. “phagy” which is an ingestion process. Since we know that gases can be moved it follows that the implicit relation is a relation between a process and a moving object.

["cardio"]+["megaly"]=["large","size","of","the","heart"]

explanation:

The compound consists of one entity-noun i.e. “cardio”, which is a body part and one adjective (property) i.e. “megaly” which concerns to size. Since we know that body parts have size it follows that the implicit relation is an attributive relation.

3.5. A subsystem for hypernym extraction from term definitions

The fifth subsystem is based on a method of hypernym extraction from term definitions. The Extraction of the hypernyms of the definitions is based on the use of a number of patterns with which the appropriate noun denoting the hypernym is specified. Some of such patterns are:

"det adjective noun".

"det adjective adjective noun".

"adjective conjunction adjective noun".

"verb det noun".

Text Tagging

The function of text tagging is used for the preparation of our texts before subjecting them to syntactic analysis. For the purposes of the present system the main application was the tagging of the definitions of terms. The definitions of terms found in the machine readable dictionaries used need to be tagged with part of speech tags so that the patterns defined in the next phase of processing for the extraction of hypernyms can be applied to them.

The input file for this module of the system will be obtained from Machine Readable Dictionaries after applying some pre processing programs that remove unnecessary information from the original electronic dictionaries. The tagging of the input file obtained will be supported by a morphological dictionary containing all the words of the definitions. This morphological dictionary will be compiled by hand and enriched with new entries whenever our tagging program detects a word in the input file not contained in the morphological dictionary. The part of speech tags used are:

a for adjective

ad for adverb

c for conjunction

det for deterniner

n for noun

p for present participle or gerund

pp for past participle

r for relative pronoun

v for verb

The input file for this module of the system is obtained from the Machine Readable Dictionaries use after applying some pre processing programs that removed unnecessary information from the original electronic dictionaries. The tagging of the input file obtained was supported by a morphological dictionary containing all the words of the definitions. This morphological dictionary was compiled by hand enriched with new entries whenever our tagging program detected a word in the input file not contained in the morphological dictionary. For illustrative purposes we present below the first ten preprocessed word definitions of the input and the corresponding output of the tagging program showing the first four tagged words of each of the input definition. The number of words shown is efficient for determining the hypernym of each term as will be explained in the next section.

Examples from the Input File

Abortion - termination of a pregnancy; can occur because of natural causes (called a miscarriage) or be a medical intervention

Abscess - an accumulation of pus in a body tissue, usually caused by a bacterial infection

Acidosis - a condition marked by abnormally high acid levels in the blood, associated with some forms of diabetes, lung disease, and severe kidney disease

Acne - a skin condition characterized by inflamed, pus - filled areas that occur on the skin's surface, most commonly occurring during adolescence

Afterbirth - the placenta and membranes that are eliminated from the woman's uterus following the birth of a child

Afterpains - normal contractions of the uterus after childbirth that usually occur for the first few days after delivery

Airways - the passageways that air moves through while traveling in and out of the lungs during breathing

Albinism - a condition in which people are born with insufficient amounts of the pigment melanin, which is responsible for hair, skin, and eye color

Allergen - a substance that causes an allergic reaction

Allergy - a negative reaction to a substance that in most people causes no reaction

Amniocentesis - a procedure in which a small amount of amniotic fluid is removed from the mother’s womb in order to detect abnormalities of the fetus

Analgesic - a drug that relieves pain, such as aspirin or acetaminophen

Androgen - a hormone (such as testosterone) that causes development of male characteristics and sex organs

Anemia - a condition in which the blood does not contain enough hemoglobin, the compound that carries oxygen from the lungs to other parts of the body

Anencephaly - a fatal birth defect in which the brain and spinal cord have failed to develop, resulting in the absence of a portion of the skull and brain

Anesthesia - a loss of sensation in a certain part of the body or throughout the body

Part of the Corresponding Entries of the Output File with Tagged Definitions

Abortion = a det termination n of prep a det

Abscess = a det accumulation n of prep pus n

Acidosis = a det condition n marked pp by prep

Acne = a det skin n condition n characterized pp

Afterbirth = the det placenta n and conj membranes n

Afterpains = normal a contractions n of prep the det

Airways = the det passageways n that r air n

Albinism = a det condition n in prep which r

Allergen = a det substance n that r causes v

Allergy = a det negative a reaction n to prep

Amniocentesis = a det procedure n in prep which r

Analgesic = a det drug n that r relieves v

Androgen = a det hormone n such ad as c

Anemia = a det condition n in prep which r

Anencephaly = a det fatal a birth n defect n

Anesthesia = a det loss n of prep sensation n

Hypernym Extraction from Lexical Definitions

The Extraction of the hypernyms of the definitions is based on the use of a number of patterns with which the appropriate noun denoting the hypernym is specified. Some of the patterns used are:

"adjective noun".

"adjective conjunction adjective noun".

"adjective adjective noun".

"det noun".

"det adjective adjective noun".

"det adjective noun noun".

"det adjective noun".

"det det adjective noun".

"det noun noun".

"det noun relative noun".

"det noun". "noun noun".

"noun conjunction noun".

"noun noun".

"noun relative adjective noun".

"noun relative det noun".

"noun relative noun".

"noun".

"verb det noun".

In cases where a term is a noun and its tagged definition matches a pattern that has only one noun this noun is automatically chosen by the program as the appropriate hypernym of the term. In cases where the tagged definition of the term matches with a pattern containing more than one noun some constaining rules are used based on medical knowledge given to the system that forbid certain categories of nouns to be used as hypernyms. An example of such categories of nouns are body parts because we define strictly the concept of hypernym as excluding meronomic knowledge.

4. CONCLUSIONS

In this paper we described the historical development and the architecture and use of a modern bilingual ontology-based system for Medical Informatics with a friendly natural language interface that accomplishes a number of functions related to the processing of Medical Texts. The functions of the system are performed by the subsystems for a friendly natural language user interface, for the resolution of the ellipsis of the agent of an action, for the discovery of implicit semantic relations between the elements of a noun phrase, for the processing of compound medical terms and for hypernym extraction from term definitions. The main novelty of this subsystem is the generation of explanations that support medical terminology management. We found also that it was possible to create a taxonomy of medical terms by automatic processing of the definitions of the terms. We also demonstrated the possibility of the automatic synthesis of the definitions of compound terms by analyzing them automatically into their component terms and combining automatically the definitions of the components.

REFERENCES

1. Kontos, J., Malagardi I. (1998). Information and Knowledge Extraction from Texts. Health Telematics Education Conference. Athens.

2. Kontos, J., Malagardi I. (1999). Information Extraction and Knowledge Acquisition from Texts Using Bilingual Question–Answering. Journal of Intelligent and Robotic Systems 26 (2): October. Kluwer Academic Publishers. pp. 103-122.

3. Kontos, J., Malagardi, I. (2000a). Greek Computational Dictionary. Project Report of the Postgraduate Course Project in Health Informatics. Coordinated by the Nursing Department of the University of Athens (in Greek).

4. Kontos, J., et al. (2000b). ARISTA Generative Lexicon for Compound Greek Medical Terms. LREC 2000, Vol. II, pp. 1075-1708.

5. Kontos, J. et al. (2002c). ARISTA Causal Knowledge Discovery from Texts. Proceedings of the 5th International Conference DS 2002. Luebeck, Germany, November 2002, Springer Verlag. pp. 348-355.

6. Kontos, J. et al. (2004). The AROMA System for Intelligent Text Mining. HERMIS International Journal (to appear).

7. Rossi Mori, A., et al. (1997) An Ontological Analysis of Surgical Deeds, in C. Garbay, et al. (eds), Proceedings of Artificial Intelligence in Europe (AIME ’97), Berlin: Springer Verlag.

8. Schulze-Kremer, S., (1997). Adding Semantics to Genome Databases: Towards an Ontology for Molecular Biology. Proceedings of the Fifth International Conference on Intelligent Systems for Molecular Biology. T. Gaasterland, et al. (eds.), Halkidiki, Greece.

9. Schubert, R. (2001). Bones, Holes, and Scales – On the Need for a Spatial Ontology for Anatomy, Manuscript, University of Hamburg.

10. Smith, B. (2001). Ontology and Information Systems. Paper based upon work supported by the National Science Foundation under grant number BCS-9975557 (Ontology and Geographic Categories).

11. Kontos, J. (1972). Program Generators Controlled by Natural Language, Calcolo, Supplemento al Vol. IX.

12. McCarthy, J. (1980). Circumscription – A Form of Non-Monotonic Reasoning, Artificial Intelligence, 5: 13, 27–39.

13. Hayes, P., J. (1979) The Naïve Physics Manifesto, in D. Michie (ed.), Expert Systems in the Micro-Electronic Age, Edinburgh: Edinburgh University Press, 242–70, cited according to the reprint in Margaret Boden (ed.), (1990) The Philosophy of Artificial Intelligence, Oxford: Oxford University Press, pp.171–205.

14. Hayes, P., J. (1985a) The Second Naive Physics Manifesto, in Hobbs and Moore (eds.), 1–36.

15. Hayes, P., J. (1985b) Naïve Physics I: Ontology for Liquids, in Hobbs and Moore (eds.), pp.71–108.

16. Sowa, J., F. (1984). Conceptual Structures. Information Processing in Mind and Machine, Reading, MA: Addison Wesley.

17. Alexander, J. H. et al. (1986). Knowledge Level Engineering: Ontological Analysis. Proceedings of AAAI–86. Proceedings of the 5th National Conference on Artificial Intelligence, Los Altos: Morgan Kaufmann Publishers, pp.963–968.

18. Gruber, T. R. (1993). A Translation Approach to Portable Ontology Specifications. Knowledge Acquisition, 5, pp. 199–220.

19. Genesereth, M., R., Nilsson, L. (1987). Logical Foundation of Artificial Intelligence, Los Altos, California: Morgan Kaufmann.

20. Gruber, T. R. (1995). Toward Principles for the Design of Ontologies Used for Knowledge Sharing. International Journal of Human and Computer Studies, 43(5/6), pp.907–928.

21. Guarino, N. (1995). Formal Ontology, Conceptual Analysis and Knowledge Representation. International Journal of Human-Computer Studies, 43, 625–640.

22. Guarino, N. (1999) The Role of Identity Conditions in Ontology Design, in C. Freksa and D. M. Mark (eds.), Spatial Information Theory: Cognitive and Computational Foundations of Geographic Information Science, Berlin/New York: Springer Verlag, pp. 221–234.

23. Guarino, N., Welty, C. (2000) A Formal Ontology of Properties, in R. Dieng and O. Corby (eds.), Knowledge Engineering and Knowledge Management: Methods, Models and Tools. 12th International Conference (EKAW 2000), Berlin/New York: Springer: pp. 97–112.

24. Malagardi, I. (1996). Computer Determination of Relations between the Elements in Noun Phrases of Sublanguages. 17th annual meeting of the Department of Linguistics. Aristotle Univ. of Thessaloniki. pp.508-520.