Dictionary and Ontology of Education and Science Terms

IRENA MARKIEVICZ, ERIKA RIMKUTĖ

Vytauto Didžiojo universitetas

KEYWORDS: corpus, education and science terms, ontology, dictionary

INTRODUCTION

In 2010–2012, a project funded by the Research Council of Lithuania, Automatic Identification of Education and Science Terms (ŠIMTAI 2), was carried out at the Centre for Computational Linguistics of Vytautas Magnus University. The main aim of this project was to test corpus linguistics methods in order to enable the automatic or semi-automatic recognition of terms, automate the extraction from the corpus of information needed to formulate term definitions, and thus facilitate the work of terminologists. In addition to the methodologies, the most important results of the research are a specialized education and science (hereinafter ES) corpus (4 million words in size), the Dictionary of Education and Science Terms prepared on the basis of this specialized corpus, and the Ontology of Education and Science Terms.

This article presents the ES term dictionary and the ES term ontology. It describes the structure of the dictionary and how it differs from other term dictionaries. Greater attention is devoted to the ES term ontology. The article explains how ontologies are defined in different fields of science, what they are needed for, and the methods by which they are created. The ES term ontology is discussed in detail; its structure is presented, examples are provided, and the article explains how this resource can be used.

Both works—the ES term dictionary and the ES term ontology—were prepared on the basis of the same material, namely a specialized ES corpus, using the methodology of corpus and computational linguistics.

This work aims to present the possibilities offered by corpus and computational linguistics in the field of terminology and to draw the attention of terminologists and compilers of term dictionaries or databases to the fact that term management can be automated, at least in part, and that prescriptive and descriptive approaches to term analysis can be reconciled.

56

Irena Markievicz, Erika Rimkutė

Švietimo ir mokslo terminų žodynas ir ontologija

STAGES IN COMPILING THE DICTIONARY OF EDUCATION AND SCIENCE TERMS

The terms in the Education and Science (E&S) terminology dictionary were identified in a specialized corpus using linguistic methods (for more details, see Grigonytė et al. 2011; Rimkutė 2012). The dictionary was compiled over two years, i.e., for the duration of the above-mentioned project, in the following stages:

1) a specialized E&S corpus of 4 million words was compiled;

2) the corpus was automatically morphologically annotated;

3) a tool for automatically extracting terms, based on linguistic rules, was developed;

4) criteria for selecting potential terms were defined;

5) approximately 11,000 potential terms, ranging from one to fifteen words, were identified;

6) additional criteria for selecting potential terms were applied¹;

7) 780 E&S terms (1–5 words) were identified;

8) a corpus-based methodology for defining terms was developed;

9) linguistic resources required to establish term definitions were prepared;

10) terms from 300 selected groups were defined;

11) English, German, and French equivalents of the defined terms were provided.

Undoubtedly, compiling any terminology dictionary requires considerable specialist working time and financial resources. This article presents the E&S terminology dictionary, which, although small—it contains 224 terms (for more details, see the next section)—is methodologically very complex. As the stages of compiling the dictionary described above show, even with a specialized corpus in the relevant field², special tools were needed³ to automatically recognize and identify terms.

¹ Terms consisting of no more than seven words were analyzed; frequency was also taken into account: the shorter the term, the more frequently it had to be used. For example, only one-word terms used at least 20 times were analyzed, while seven-word terms had to occur at least twice.

² Corpus compilation is one of the most important and lengthy stages of the entire study, since compiling a representative corpus in which terms could be identified required carefully selecting texts, balancing them by genre, and processing them technically (converting PDF files into text files, changing the encoding, etc.).

³ In this project, a tool adapted to Lithuanian for automatically identifying terms was developed because, as previous experiments show (see Grigonytė et al. 2011), due to differences in language structure, it is almost impossible to adapt term recognition tools designed for other languages to Lithuanian. The tool used linguistic rules: it searched for terms with certain grammatical structures, e.g., adjective + noun, noun in the genitive + noun, participle + noun, etc. Researchers of Latvian use similar methods to identify terms (see Gornostay et al. 2012; Krugļevskis et al. 2005). According to them, identifying terms using linguistic methods (morphosyntactic analysis) is more suitable than statistical methods suited to analytical languages.

Terminologija | 2013 | 20

57

It was also difficult to select the terms related to education and science from all the automatically identified meaningful phrases with grammatically correct structures. The difficulty arose because as many as 11,000 so-called possible terms were identified in a corpus of 4 million words. In the next stage of the research and dictionary compilation, it was important to identify the education and science terms from among all the phrases (this problem is discussed in Rimkutė 2012).

It should be noted that the presented dictionary of education and science terms was compiled on the basis of a corpus. This means that the method used was not the traditional, still widespread in Lithuania, approach to terminology management based on terminologists’ expertise, sometimes quite subjective and prescriptive; instead, it was the descriptive approach to term analysis and definition, considered superior in contemporary terminology and based on corpus data. Corpora are very useful for compiling dictionaries, because they make it possible to identify the most typical and frequent patterns of term use, which are important for defining meaning and selecting relevant examples.

Because of the complex methodology for defining terms, it was decided to provide detailed definitions for 300 terms. Terms from certain groups were selected: terms with the word academic (e.g., academic freedom, academic ethics, academic hour), with the word work (e.g., degree thesis, homework, independent work), with the word education (e.g., university education, general education, higher education), with the word education (e.g., education provider, adult education, non-formal education), and so on. Several frequently used one-word terms were also defined: student, researcher, university, etc.

As in other terminology dictionaries, most terms in the dictionary of education and science terms are accompanied by equivalents in English, German, and French. These were collected from approved sources in the relevant field, documents, and specialized terminology databases: Anglų kalbos švietimo terminų žodyno (Glossary of Educational Terms), Aiškinamojo su studijomis susijusių terminų žodyno, Der Europäische Qualifikationsrahmen für lebenslanges Lernen, EuroVoc (the multilingual thesaurus of the European Union), Lietuvos Respublikos terminų banko, etc.

The terminology of education in Great Britain, France, and Germany was mainly used as a reference; however, in the context of the systems of those countries, the meanings of many translation equivalents are not exactly the same as those of the original terms used in the Lithuanian education and science sectors. The foreign-language equivalents of terms provided in the dictionary apply to the science and education system of the Republic of Lithuania and may not always (or may only partially) apply to describing another country’s science and education system.

58

Irena Markievicz, Erika Rimkutė

Švietimo ir mokslo terminų žodynas ir ontologija

STRUCTURE OF THE DICTIONARY OF EDUCATION AND SCIENCE TERMS

The dictionary of education and science terms is freely available online at http://daukantas.vdu.lt/moksliniai-terminai/.. It contains 224 terms, although the previous section stated that 300 terms had been defined. The numbers of defined terms and terms included in the dictionary differ because it was observed that some terms (e.g., qualification level / tier; academic / scientific degree; research / scientific output; postgraduate / master’s studies; student learning / workload; state / national research institute) are used as synonyms and have similar structures, so these terms were combined.

The dictionary of education and science terms contains the components required in all terminology dictionaries, namely the term itself and its definition. As mentioned, equivalents in English, German, and French are provided. Subject areas (e.g., higher education management, higher education policy, higher education administration) and abbreviations for some terms are also included.

The dictionary differs from other terminology dictionaries or databases in that it also lists all inflected forms of terms with their frequencies, as well as the overall frequency of each term in the education and science corpus. At least two usage examples from the education and science corpus can be found for every term (the examples have not been edited or altered, only occasionally shortened), along with left and right collocates⁴ and a concordance⁵.

There was not enough information from the corpus to define every term. Asterisks are marked next to terms for which there was insufficient information to formulate definitions (e.g., education provider, student voucher, basic education).

Since the terms were extracted from a corpus, their frequency of occurrence was known, so a conventionally marked frequency relative to other terms is indicated for all terms:

★★★ is shown next to the most frequent terms, which occur up to 500 times in the education and science corpus;

★★☆ is shown next to less frequent terms, which occur 100–499 times in the corpus;

⁴ These are the contextual partners of the term being analyzed, identified statistically (by frequency), which often occur next to the term on its left and right.

⁵ These are instances of all grammatical forms of the selected term occurring in the corpus, presented in a one-line context (a concordance line usually covers more than one sentence). A concordance is generally analyzed by first reading it vertically, since only then does typical term usage become apparent. The term being analyzed is placed in the middle of the concordance line and is highlighted. Approximately 50 characters of the concordance extract are shown to the left and right of the term being analyzed. To view a longer concordance extract (approximately 150 characters to the left and right), click the plus sign at the end of the concordance line.

Terminologija | 2013 | 20

59

★☆☆ is shown next to terms used 10–99 times in the corpus;

☆☆☆ is shown next to rare terms used fewer than 10 times in the corpus.

The article for the term bachelor’s studies is presented below as an illustration of the terminology dictionary of education and science.

BACHELOR’S STUDIES ★☆☆

Definition

First-cycle higher education studies that provide a general university education, the foundations of knowledge in the chosen field of study, and a bachelor’s qualification degree.

Examples

At universities in the United States, it is common for opportunities to be provided for more gifted students to combine master’s studies with doctoral studies, or more precisely, to proceed to doctoral studies immediately after bachelor’s studies (this partly explains why master’s and doctoral studies are combined under one term—graduate studies, which are conducted in what is known as graduate school) and to earn a PhD degree after 8 years of university study.

Recommended duration in combination with bachelor’s studies – 5 years

Bachelor’s studies (first cycle) usually last 4 years (160 credits, full-time study).

Bachelor’s studies, master’s studies, and doctoral studies are completed by defending, respectively, a bachelor’s thesis, a master’s thesis, and a dissertation.

Bachelor’s studies are first-cycle studies that provide a general university education and the foundations of knowledge in the chosen field of study.

Frequency

bachelor’s studies 21

bachelor’s studies 6

in bachelor’s studies 3

bachelor’s studies 7

of bachelor’s studies 14

Overall frequency: 51

Equivalents

English: bachelor’s studies

German: Bachelorstudium

French: études de licence

Subject area

higher education administration

60

Irena Markievicz, Erika Rimkutė

Švietimo ir mokslo terminų žodynas ir ontologija

Collocates

Left collocates

Right collocates

PO

bachelor's studies

SU

bachelor's studies

ABBREVIATE

bachelor's studies

SOFTWARE

for bachelor's studies

bachelor's studies

FIRST CYCLE

bachelor's studies

MASTER'S AND DOCTORAL STUDIES

Concordance⁶

[1]

work or continue and complete

bachelor’s studies.

Representatives of the higher education institution issued a statement,

[2]

in countries offering such bachelor’s programs.

Bachelor’s studies

in the autumn semester of the fourth year, while master’s studies

[3]

theology program – 5 years); to

bachelor’s studies

1,870 students were admitted, while in 2009 the universities

[4]

universities decided to shorten to 3 and 3.5 years

bachelor’s studies,

states that the Law on Science and Studies

[5]

universities decided to shorten to 3 and 3.5 years

bachelor’s studies.

In the United States, bachelor’s degree holders, seeking to begin master’s

[6]

commission requirements: one must have completed

bachelor’s studies,

if applying to a research-oriented programme,

[7]

m. Universities were prepared to shorten bachelor’s studies by up to 3 years

bachelor’s studies.

Several universities were prepared to offer programmes lasting up to 3 and 3.5 years

[8]

credits (4.5-year) bachelor’s studies.

bachelor’s studies,

lasting 4 years (theology studies at Catholic

[9]

studies that do not award [a degree]. The recommended period, together with

bachelor’s studies

– 5 years (3 + 2; 3.5 + 1.5; 4 + 1). Bachelor’s

[10]

Asked to compare these studies with his own

bachelor’s studies

in the United States, the foreigner noted that things are much easier here.

⁶ Only the first ten lines of the concordance are presented here. Clicking the plus sign on the website displays a longer concordance.

Terminologija | 2013 | 20

61

Several thousand concordance lines are provided for some terms. Reviewing this volume of information takes a great deal of time, and not all of the data is relevant to everyone. For those wishing to learn more about a term’s usage and the context in which a selected term is used, perhaps to formulate another definition, select useful examples, or form an objective opinion, the concordance is highly beneficial.

THE CONCEPT OF ONTOLOGIES AND REASONS FOR THEIR CREATION

This and the following sections discuss the ontology of education and science (ES) terms, the principles underlying its development, and the possible uses of this and other ontologies. For clarity, several visualized classifications and results from the ES terms ontology are presented.

Ontologies are formal tools for conceptualizing knowledge, summarizing knowledge described in natural language as a structured, hierarchical model of related concepts. Problems with the quality of information processing and the ambiguity of terms create a need to develop structured term bases that computer programs can use to process human natural language (natural language processing) and interpret its semantics. Ontologies are developed for this reason.

A person who knows a particular language easily understands the meaning of a natural-language text presented to them, whereas a computer program interprets the same information as a set of text symbols and associated metadata. Ontologies are the link that transforms information written in natural language into a conceptual knowledge model that information retrieval and processing robots can readily interpret.

Ontology development is an interdisciplinary process encompassing philosophy, linguistics, and computer science (see Grigonytė 2010 for more details). In philosophy, ontology is understood as the discipline that studies being and existence. Linguists supplement this definition of ontology: in linguistics, an ontology is a network of individual concepts and the terms that describe them. In computer science, an ontology is a logical and formal representation of reality.

The largest and most readily accessible source of natural-language texts is the internet; therefore, ontologies are often understood as one of the tools of the Semantic Web and semantic search. The W3C organization, which develops standards for websites, describes an ontology as a set of terms that describes the knowledge of a particular domain: “ontologies are used by people, databases, and applications that seek to share information about a particular domain

62 Irena Markievicz, Erika Rimkutė | Švietimo ir mokslo terminų žodynas ir ontologija

information <...>. Ontologies link concepts in a particular domain, adapted for computer programs, and the relationships between them” (W3C).

The Ontology of Education and Science Terms (hereafter, the ES terms ontology) was developed to formalize the set of ES terms created in the aforementioned introductory project and to supplement it with semantic relations that applications can interpret. The main tasks in developing an ontology are to describe the terms and concepts of the selected domain comprehensively and to specify their meanings formally. These tasks are important for addressing the problem of term ambiguity and developing a structural terminology model for the ES domain. The ES terms ontology was created not only as a taxonomy of ES terms but also as one possible resource for semantic search.

The ES terms ontology is available online at http://meta.vdu.lt/webprotege/, and is visualized using the WebProtege tool, which allows users to view the hierarchical class tree, their attributes, and examples of concept instances. By opening or hiding tabs and panels, users can arrange a convenient graphical interface. The Chrome or Mozilla Firefox browsers are recommended. To display the relationships between terms more clearly, it is recommended to use Protege 4.3 with the OntoGraf plugin.

SELECTION OF A METHODOLOGY FOR CREATING AN ONTOLOGY OF EDUCATION AND SCIENCE TERMINOLOGY

The generalized concept of ontology development is described by the following process metamodel (see Cimiano 2006 for more details): M={D, LA, T, S, C, TR}, in which D denotes document collection, LA—linguistic annotation, T—term selection, S—synonym identification, C—linking terms and their synonyms into concepts, and TR—defining semantic relations between concepts.

Different ontology development methodologies employ various development strategies: some methodologies focus more on the ontology life cycle and prototype development (Cyc, TOVE, Methontology, On-To-Knowledge), while others place greater emphasis on using ontologies in applications (KAKTUS). Some methods develop ontologies from collected text corpora (TOVE, On-To-Knowledge), while others draw on large general-purpose knowledge ontologies (SENSUS, Cyc) (see Corcho et al. 2003 for more details).

Ontologies are most simply developed in the following stages: the knowledge domain is selected, information is processed (concepts, terms, and constraints are defined),

Terminologija | 2013 | 20 63

methods for extending the ontology and evaluation criteria are specified, and the ontology documentation is prepared (see Uschold et al. 1996 for more details). When ontologies are developed using this method, ontology prototypes are not envisaged, nor are ontologies intended to be used as additional information sources. This methodology has been used to develop several business ontologies, but it does not specify how they can be used in applications.

The On-To-Knowledge project proposed a methodology focused primarily on knowledge management and on the potential uses of the future ontology in information systems (see Staab et al. 2001 for more details). A feasibility study was conducted, and information sources, potential ontology users, and use cases were identified.

Experts in the selected domain participated in the development of the On-To-Knowledge ontology; information gathered from texts and other knowledge bases was processed. Ontology quality was evaluated by checking whether it met the requirements and could be applied. All these stages in ontology development and refinement can serve as recommendations for creating other ontologies.

The methodologies presented focus primarily on methods for developing ontologies from textual information sources. The SENSUS ontology methodology proposes using general-knowledge ontologies in the development of a domain-specific ontology (see Gomez-Perez et al. 2004 for more details). The methodology provides for grouping terms from the OntoSaurus knowledge base into a concept tree. When attempting to link two concept classes, a general class of concept (English: root class) is sought.

The methodology for developing an ontology of education and science (ES) terminology was selected with consideration of the available information resources and the applicability of language technologies and ontology development tools. It was decided to draw on the experience accumulated in the Methontology project (see Mariano et al. 2004; Gomez-Perez et al. 2004; Corcho et al. 2003 for more details).

The application of Methontology to developing an ontology of ES terminology, together with the resources for ontology development, is shown in Figure 1. The specification states the purpose and scope of the ontology. The conceptual ontology model is developed with regard to the domain refined during specification and is subsequently formalized—that is, converted into a formal ontology model. During the development process, the formal model is implemented according to the requirements of the selected ontology development language. Even after an ontology has been created, it is often modified, expanded, or updated.

64

Irena Markievicz, Erika Rimkutė

Švietimo ir mokslo terminų žodynas ir ontologija

Fig. 1. Processes and resources for developing the terminology ontology of education and science (based on Methontology)

Applying the Methontology model, a terminology glossary and concept taxonomy were created, relationships between individual ontology objects were identified, and a database of example attributes, classes, abstract classes, and their attributes and relationships was prepared. It specifies the circumstances in which a selected concept exists, the other concepts with which it is related, and also describes the concept instances.

Ontologies describe different things within the same specific domain: persons (e.g., rector, student), places (e.g., university), processes (e.g., teaching, lecturing), process outcomes (e.g., work of art, bachelor’s thesis), etc. In the ontology, all of these are named as concepts (English: concept), which may exist as independent units of ontological information or may be linked into concept classes (English: class).

A concept taxonomy (English: concepts taxonomy)—a systematic classification of concepts according to the selected conceptual ontology model—is usually created from a terminology glossary (English: glossary of terms), which in this case is understood as a formal list of terms and their definitions (see Uschold et al. 2004 for more details).

In classical taxonomy, a class is defined as “a group of similar entities that differs from other groups of individuals” (see Mayr 1987 for more details). Each ontology class is characterized by a set of properties (English: property) unique to it and their restrictions (English: restriction), which are inherited by all instances (English: instance, individual) of a given class. An instance is a concrete example of the selected concept (e.g., Zigmas Lydeka is an instance of the concept rector).

Terminologija | 2013 | 20 65

The terms in the education and science terminology ontology were identified in a specialized corpus using linguistic methods (see Grigonytė et al. 2011; Rimkutė 2012 for more details). The ontology concepts and concept classes were developed using a mixed bottom-up (English: down-top strategy) and top-down (English: top-down strategy) strategy (see Uschold et al. 1996 for more details). When applying the bottom-up strategy, the most specific concepts are identified first and are later replaced with more abstract ones. When applying the top-down strategy, abstract concepts are defined first and are later refined. When applying a mixed ontology-building strategy, the initial concepts are examined first and, depending on requirements, are either refined or generalized.

An expert in the education and science domain actively participated in this stage of developing the ontology terminology taxonomy. The expert grouped the terms by subject area and created an initial conceptual model of the ontology.

When developing a concept taxonomy, it is recommended to pay particular attention to the following stages: compiling a list of terms and selecting the necessary terms from it, creating a concept model, preparing definitions, and evaluating the terms (ISO 10241: 1992).

STRUCTURE OF THE ONTOLOGY OF EDUCATION AND SCIENCE TERMS

One of the challenges in the ontology development process is to create a high-quality data structure that is universal and adaptable to different information-processing systems, and compatible with other knowledge sources. The main element of an ontology is a thing (English thing), which abstracts all the concepts in the ontology. The necessity of this element is explained by the philosophical definition of ontology as a science that studies being. An ontology formalizes all the knowledge in a particular domain, so an essential element is an abstract class that generalizes all the concepts of a specialized domain.

Concepts, terms, and axioms are the principal conceptual elements of any ontology. It is important to distinguish between the meanings of a term and a concept. In the LST ISO 1087-1 standard (2005), concept is defined as “[a] unit of knowledge created by a unique combination of characteristics.” A concept is independent of language and can be influenced only by context. A term is an identifier of a concept, that is, “[a] verbal designation of a general concept in a special subject field” (LST ISO 1087-1: 2005). Axioms describe the constraints of a specific domain and establish a hierarchical structure of the concepts in the ontology. Formal structural ontology objects include concepts, concept classes, instances of classes and concepts, properties of classes, concepts and instances, and relations and constraints linking all ontology objects.

66

Irena Markievicz, Erika Rimkutė

Švietimo ir mokslo terminų žodynas ir ontologija

Fig. 2. Structure of the education and science terms ontology by business process model domains

On the expert’s recommendation, all terms were grouped into three abstract classes—input, process, and output (see Fig. 2)—characteristic of business process models (see ISO 9001: 2000 for further details). The Input class specifies the requirements, resources, and data of a system or a particular activity. Accordingly, the term groups infrastructure, institution, and participant are assigned to this category. Business processes are divided into core processes (the Activity class), which create added value, and support processes (the Administration class), which strengthen the operation of core processes and provide infrastructure for the processes being carried out. Output summarizes process effectiveness and efficiency—it indicates what results could be achieved, what results were achieved, and what resources were used to achieve them (the research output, vocational training output, study output, and artistic output classes). Each abstract domain class may have additional domain-specific subclasses (as shown in Fig. 3, the participant class has administrator as a subclass, which is further subdivided). The education and science terms ontology contains 29 abstract classes in total. Their main purpose is to create a general conceptual model of the ontology according to the aforementioned top-down strategy.

Terminologija | 2013 | 20

67

Fig. 3. General concepts of the Participant class and abstract subclasses in the education and science terms ontology

The education and science terms ontology consists of 465 general concepts⁷, whose hierarchy was created using definitional-context recognition patterns (see Bielinskienė et al. 2012 for further details). Definitional contexts partly or fully correspond to term definitions and are formed from hypernymic concepts. For example, the concept student is defined as a person studying at a higher education institution under a degree programme. A foreign student is a student who has come from another country to study at a higher education institution in a particular country, either for a full degree programme or for part of a programme.

The ontology of science terms specifies the following taxonomic relations:

– subclass-of (English subclass-of) – e.g., master’s student is a subclass of the student class;

– disjoint decomposition (English disjoint-decomposition) – e.g., study output cannot be both a bachelor’s thesis and a dissertation at the same time;

– exhaustive decomposition (English exhaustive-decomposition) – e.g., the lecturer class includes the subclasses associate professor, lecturer, professor, and others. If an object is a lecturer, it must belong to at least one of the listed subclasses;

– partition (English partition) – e.g., the concept study semester comprises the concepts spring semester and autumn semester. A study semester must belong to one of the listed subclasses.

⁷ The ontology also includes terms that are not found in the education and science terms dictionary, that is, terms that were not fully defined.

68

Irena Markievicz, Erika Rimkutė

Švietimo ir mokslo terminų žodynas ir ontologija

Figure 4. Taxonomic hierarchy of the concept student in the terminology ontology of education and science

Since the main criterion for ontology quality is the unambiguousness of terms, considerable attention was paid during the development of the education and science terminology ontology to ensuring that the set of concept properties was compiled with care. Descriptive context recognition patterns are suitable for determining the set of properties of classes, concepts, and entities (see Bielinskienė et al. 2012 for more details). Descriptive contexts specify the nature of the concept being analyzed and its relationship to other parts of the system.

Figure 5. Possible applications of the attribute active in the education and science terminology ontology⁸

The same property can be used to describe several different concepts. For example, the property active applies to four concepts: practical work, practical training, practical session, and exercises (see Figure 5). It is therefore important to emphasize that a concept can be described adequately only by a particular, unique combination of properties.

⁸ Using the Protege 4.3 ontology development and visualization tool, it was not possible to use Lithuanian; therefore, structural ontology elements (e.g., ObjectProperty, Domain) are presented in English.

Terminologija | 2013 | 20

69

Figure 6. Set of properties and their restrictions for the concept of an academic hour in the education and science terminology ontology

Each property may have specified restrictions or relationships with other concepts or concept classes. For example, an academic hour is defined as a unit of the duration of contact work between teachers and students, equal to 45 minutes. The formal, ontological definition of this concept specifies the following set of properties: an academic hour is a subclass of a process; the participants are students and a teacher; it is a contact process with a duration, defined by the constant 45 minutes (see Figure 6).

Entities—instances of specific classes and concepts—inherit the properties of classes and concepts, their restrictions, and their relationships with other properties, and are characterized by

Figure 7. Tree of relationships between concepts and entities in the education and science terminology ontology

70

Irena Markievicz, Erika Rimkutė | Švietimo ir mokslo terminų žodynas ir ontologija

with the same properties. For example, if an ontology specifies that the rector leads a university, then the instance of this concept, Zigmas Lydeka, leads VDU, i.e. an instance of the concept university (see Fig. 7).

The ontology of education and science terminology was created in OWL using the Protege 4.3 ontology development tool and visualized using the WebProtege tool or the OntoGraf plugin⁹.

EVALUATION AND APPLICATION OF THE ONTOLOGY OF EDUCATION AND SCIENCE TERMINOLOGY

At present, the ES terminology ontology contains 29 abstract concept classes and 465 general concepts, 87 properties of concepts, concept classes and their instances, 141 instances, and 296 relationships among ontology concept classes, concepts and instances.

Formally, ontology complexity can be assessed using description logic (English description logic) metrics (for more details, see Corcho et al. 2003; Baader 2003). The ontology of educational science terminology is assessed at the ALCHON(D) complexity level, where:

– AL – an attribute language; it can be used to connect concepts and write universal restrictions for them;

– C – indicates that concept relationships can be negated;

– H – indicates that a role hierarchy can be created;

– O – cardinality restrictions (relationships: exactly one, one of many);

– N – number restrictions (relationships: at most, at least);

– (D) – data type properties.

In addition to the ontology’s vocabulary, taxonomy and semantic relationships, its potential for use and application is evaluated. The selected OWL ontology development language is suitable for making semantic inferences, extending the ontology with new concepts, or linking ontologies from several different domains. The structural terminology model for the ES domain was prepared so that each ontology concept is described by a unique set of properties and semantic relationships—a solution that helps address the problem of terminological polysemy. This resource was created not only as a taxonomy of terms in the ES domain, but also as one possible tool for semantic search and natural language processing.

⁹ More information: http://protege.stanford.edu.

Terminologija | 2013 | 20

71

To date, much attention has been devoted to how ontologies can be applied to the computerization of natural language, systematizing it and making it understandable to computer programs. Perhaps at first glance it is not entirely clear how ontologies can be useful to linguists or specialists in other fields. It should be emphasized that ontologies are particularly useful to terminologists because, with the help of ontology visualization tools, the relationships between terms can be clearly seen. To illustrate terms, specific words related to those terms and included in the ontology can be selected (called individuals in the ontology).

Both the Education and Science (E&S) terminology dictionary and the ontology can help participants in the E&S field: lecturers, teachers, administrative staff, students, and school pupils, because these resources show, for example, how students are classified, what types of studies there are, what the administration of higher education institutions consists of, and so on.

CONCLUDING REMARKS

The article presents the Education and Science Terminology Dictionary and the Education and Science Terminology Ontology. Both the dictionary and the ontology have been developed further. The greatest advantage and distinctive feature of both resources is that they were prepared using corpus and computational linguistics principles, which are still little used by Lithuanian terminologists.

The E&S terminology dictionary is available online and comprises 224 terms consisting of one to five words. It differs from other terminology dictionaries in that, in addition to the usual dictionary components (the terms themselves, definitions, equivalents in foreign languages), it provides all inflected forms of each term together with their frequencies, examples from the E&S corpus, collocates, and concordances. These data are valuable because dictionary users can form a comprehensive picture of how a particular term is used and its context; they can also find important features of a term that may not be mentioned in its definition.

The E&S terminology ontology, available online, reveals the semantic relationships among E&S terms and makes this information understandable to computer programs. Ontologies can be particularly useful to terminologists who classify and describe the terms of a particular field and provide specific examples.

It is hoped that both the E&S terminology dictionary and the ontology will be methodologically important resources and encourage Lithuanian terminologists to make greater use of the opportunities offered by corpora to study, describe, and classify terms.

72

Irena Markievicz, Erika Rimkutė

Švietimo ir mokslo terminų žodynas ir ontologija

LITERATŪRA

Baader F. 2003: Appendix: description logic terminology. – The description logic handbook: theory, implementation, and applications, Cambridge: Cambridge university press.

Bielinskiene A., Boizou L., Kovalevskaite J., Utka A. 2012: Towards the Automatic Extraction of Term-defining Contexts in Lithuanian. – Human Language Technologies. The Baltic Perspective: Proceedings of the Fifth International Conference Baltic HLT 2012, Amsterdam, Berlin, Tokyo, Washington, DC: IOS Press, 18–26.

Cimiano P. 2006: Ontology Learning and Population from Text: Algorithms, Evaluation and Applications, Karlsruhe: Springer.

Corcho O., Fernández-López M., Gómez-Pérez A. 2003: Methodologies, tools and languages for building ontologies. Where is their meeting point? – Data & knowledge engineering 46(1), 41–64.

Gomez-Perez A., Corcho O., Fernandez-Lopez M. 2004: Ontological Engineering: with examples from the areas of Knowledge Management, e-Commerce and the Semantic Web, London: Springer-Verlag.

Gornostay T., Ramn A., Heid U., Morin E., Harastani R., Planas E. 2012: Terminology Extraction from Comparable Corpora for Latvian. – Human Language Technologies. The Baltic Perspective: Proceedings of the Fifth International Conference Baltic HLT 2012, Amsterdam, Berlin, Tokyo, Washington, DC: IOS Press, 66–73.

Grigonytė G. 2010: Building and Evaluating Domain Ontologies: NLP Contributions, Logos Verlag Berlin GmbH.

Grigonytė G., Rimkutė E., Utka A., Boizou L. 2011: Experiments on Lithuanian Term Extraction. – Proceedings of the 18th Nordic Conference of Computational Linguistics NODALIDA 2011. NEALT Proceedings Series 11, 82–89.

ISO 10241: 1992 International terminology standards – Preparation and layout. Geneva: ISO.

ISO 9001: 2000 Quality management systems – Requirements, Geneva: ISO.

Kruglevskis V., Vancāne I. 2005: Term Extraction from Legal Texts in Latvian. – Proceedings of the Second Baltic Conference Human Language Technologies, 155–159.

LST ISO 1087-1: 2005 Terminologijos darbas. Aiškinamasis žodynas. 1 dalis. Teorija ir taikymas (tpd ISO 1087-1: 2000).

Mariano F., Gomez-Perez A., Juristo N. 2004: Methontology: From Ontological Art Towards Ontological Engineering. – AAAI-97 Spring Symposium. Series on Ontological Engineering, Stanford: AAAI Press, 33–40.

Mayr E. 1987: The ontological status of species: scientific progress and philosophical terminology. – Biology and Philosophy 2(2), 145–166.

Rimkutė E. 2010: Lietuvių kalbos terminų automatinio nustatymo galimybės. – Kalbų studijos 16, 71–78.

Rimkutė E. 2012: Automatinis švietimo ir mokslo terminų nustatymas lingvistiniais metodais. – Terminologija 19, 54–69.

Staab S., Studer R., Schnurr H. P. 2001: Knowledge Processes and Ontologies. – Intelligent Systems 16(1), 26–34.

Uschold M., Gruninger M. 1996: Ontologies: Principles, methods and applications. – Knowledge engineering review 11(2), 93–136.

Uschold M., Gruninger M. 2004: Ontologies and semantics for seamless connectivity. – ACM SIGMod Record 33(4), 58–64.

W3C – OWL Web Ontology Language Use Cases and Requirements. Prieiga internete http://www.w3.org/TR/2002/WD-webont-req-20020307/ (žiūrėta 2013-10-14).

ONTOLOGY AND DICTIONARY OF SCIENCE AND EDUCATION TERMS

The paper presents the dictionary and the ontology of science and education terms. The proposed approach of ontology and dictionary building uses the following Natural Language Processing techniques: linguistic pattern recognition, collocation extraction, morphological analysis, POS tagging, and others.

The dictionary of science and education is available online and contains 224 one word and multiword terms (their length is up to 5 words). The dictionary contains the

Terminologija | 2013 | 20

73

necessary dictionary fields (terms, definitions, translations to other languages), as well as some additional fields such as declined forms of terms with their frequencies, examples of term usage, collocations and concordances from the special science and education corpus.

The main focus of the paper is the ontology of terms in the domain of science and education. Authors present, how different research areas define ontologies, what are the possibilities of their use, what are ontology building methods. The structure of the ontology is illustrated by multiple examples. The hierarchical structure of concepts, their properties and relationships make this information readable and understandable for knowledge-based information systems. Ontologies can be particularly useful for terminology experts, who try to classify and describe terms, concepts and their instances in a domain specific field.

Both resources can be valuable for lecturers, teachers, administration management bodies, students, etc. While using the ontology, they can define and conceptualize the processes of studying, teaching or learning.

The paper introduces the possibilities of semi-supervised term extraction with mixed prescriptive and descriptive term analysis methods, which can be useful for terminology experts, lexicographers, and creators of lexical databases.

Gauta 2013-10-29

Irena Markievicz Vytauto Didžiojo universitetas K. Donelaičio g. 52, LT-44248 Kaunas E. paštas [email protected]

Erika Rimkutė Vytauto Didžiojo universitetas K. Donelaičio g. 52, LT-44248 Kaunas E. paštas [email protected]

74

Irena Markievicz, Erika Rimkutė | Švietimo ir mokslo terminų žodynas ir ontologija