Terminology Organization in Terminology Management Systems

1
Terminology Organization in Terminology
Management Systems

• Angela Boll, Marina Kaneva, Claudia Himmler,
  Chiara Huber, Annika Meinhardt, Patrick Johnson

2
COMPILATION OF TERMINOLOGY

• the most practical way to process lexical data is
  by computer
• benefits speed, flexibility and storage capacity
  
• growing trend towards the automation of
  terminological data processing
• from now on, all aspects of terminology
  compilation, storage and retrieval will be
  assisted by or directly carried out by computers

3
PRINCIPLES OF COMPILATION

• automation fundamentally affects the compilation
  of terminology
• necessity to evolve completely new principles for
  compilation

4
PRINCIPLES OF COMPILATION

• systematic terminology compilation is now firmly
  corpus-based
• text corpora reinforce the principle that
  terminology compilation is an ongoing and
  repeated activity

5
PRINCIPLES OF COMPILATION

• many technical texts can now be preserved in or
  converted into a suitable format for
  terminological analysis
• texts which are to be processed by translators
  can be analysed and compared with current
  machine-readable terminology holdings and a
  machine-readable general dictionary in order to
  produce a listing of items not contained in
  either

6
PRINCIPLES OF COMPILATION

• running text can be used totally independently of
  user requirements
• terminology compilation is becoming increasingly
  text-oriented

7
PRINCIPLES OF COMPILATION

• the second major innovation affecting principles
  of compilation is the division which is now
  possible between
• the raw data as they are found in the corpus,
• the database which contains all the information
  that is collected in suitably structured form,
  and
• all the various subsets of information which are
  created for specific purposes and uses

8
PRINCIPLES OF COMPILATION
9
PRINCIPLES OF COMPILATION

• the terminologist now has appropriate tools which
  lift his work from a craft to a scientifically
  supported activity
• automatic processing and computer-assisted
  terminology compilation is therefore
  qualitatively superior to conventional methods
• terminologist is freed from the limitations of
  the past with respect to size of individual
  records and total quantity of records

10
PRINCIPLES OF COMPILATION

• however, there is also a danger private term
  collections of individual translators can become
  widely known
• instead, there should be only one major database
  of terminological information for each language
  community, to which all users would refer and
  contribute
• communication across all industrial and
  institutional barriers would be facilitated

11
The nature and type of terminological information

• Information for the construction of a
  terminological record is various and subject to
  changes
• This affects the nature of database system
• Information in the database must be considered
  independent of each other
• Information can be entered at different times and
  from different sources

12
The nature and type of terminological information

• Full bibliographical information for each item is
  provided separately
• Limitation of human manipulation of lexical data
  to the specific interpretative tasks the computer
  cannot perform
• Concept is explained by indication of linguistic
  forms antonyms, broader and narrower generic
  terms (refer to a whole class of terms), broader
  and narrower partitive terms (relate to a part of
  a whole)

13
The nature and type of terminological information

• Exemplification of the usage of technical terms
  example sentences (context) and usage notes
• Terms meaning is semantically more changeable
  than items of the general lexicon of a langauge

14
The nature and type of terminological information

• In conceptually-based terminological data banks
  definitions are given in one language only
• Bilingual terminology is directional and
  non-reversible gt translation equivalents cannot
  be converted into entries of the source language
• Translation equivalents do not refer to an
  authentic concept because they introduce new
  concepts

15
Methodological considerations

• Terminologists dont need to be concerned how the
  data is stored in the computer thanks to the
  modern techniques of computational linguistics
• Computer can store a multi-dimensional semantic
  network
• No physical limitation of the size of any
  non-magnetic medium
• Definitions can be as long as is necessary to
  properly define the term

16
Methodological considerations

• Terminology compilation can be distributed
  physically and temporally
• Information can be collected and stored in stages
• As long as each item of data satisfies the
  controls (e.g. bibliographical reference) gt as
  much data as available can be entered at any time
• Information can be collected on a distributed
  basis gt work can be distributed among various
  people and locations gt it is particularly
  important for the compilation of multilingual
  terminology

17
Quality of data

• Computer usage for input control and validation
  resulted in a trend to terminology of a higher
  quality
• Increased dangers of spreading terminology of low
  quality
• Increase in quality is very important
• Far-reaching effect of computerised terminology
  processing on terminology spreading

18
Quality of data

• Distinction between original source texts and
  translated texts
• Terms taken from texts in their original language
  genuine terms and as such have full validity
• Terms taken from translated texts may either be
  valid terms or translation equivalents

19
Quality of data

• Trend towards the use of genuine original texts
  for extraction of terms and contexts
• There is no exact match of concepts for many
  terms across languages
• Several possible equivalents together with
  context and usage information are needed for a
  correct choice

20
Principles of data collection

• Set of basic principles for the compilation of
  terminological data
• Certain consistency of criteria
• Sources must be stated
• Distinction between original and translated texts
• Linguistic behaviour of terms should be
  documented by contexts so that all relevant
  textual variants are covered

21
Terminological Data Banks-A Definition-

• Automated collection of vocabularies of special
  areas that serve a particular user group
• Used for large translation services
• Enhanced but still conventional glossaries
  transferred to a new medium

22
Terminological Data Banks-A Definition-

• Designed to give response to the same questions a
  good dictionary is supposed to answer
• But these questions only elicit direct responses
  from the various parts of the conventional
  dictionary

23
Terminological Data Banks-A Definition-

• Examples
• ENTRY PART QUESTION
  ANSWER
• equivalent what is the French word
  imprimante
• for laser
  printer? laser
• gender what is the gender of
  feminine
• imprimante?

24
Terminological Data Banks-A Definition-

• These responses are not sufficient for a wide
  range of dictionary users
• Answers may be ambiguous
• Full potential of a lexical database was not
  exploited by existing term banks

25
Terminological Data Banks-A Definition-

• Reasons
• Information was not unified in a suitable manner
  in order to retrieve it
• Lack of coherent structure
• Existing system failed to exploit new and
  additional techniques for ordering and
  representing the data

26
Terminological Data Banks-A Definition-

• There was an increasing demand for a system that
  allows to answer complex queries
• Example
• QUERY
  SEARCH OF FIELD
• what do you call a machine
  definition or
• that performs X?
  conceptual links

27
Terminological Data Banks-A Definition-

• a collection, stored in a computer, of
  special language vocabularies, including
  nomenclatures, standardised terms and phrases,
  together with the information required for their
  identification, which can be used as a mono- or
  multilingual dictionary for direct consultation,
  as a basis for dictionary production, as a
  control instrument for consistency of usage and
  term creation and as an ancillary tool in
  information and documentation.

28
Terminological Data Banks-A Definition-

• Term banks are supposed to be used by people
  with varying degrees of expertise and different
  purposes

29
Semantic Networks

• Complex storage of data to represent
  terminological relationships
• First developed in artificial intelligence
  research for formal representation of the human
  knowledge
• Have no intrinsic meaning they are basically
  directed graphs
• They have superficial similarity

30
Semantic Networks

• Example

31
Semantic Networks

• The relationships between concepts are expressed
  through abbreviations
• Generic relationshipis a type ofisa
• Partitive relationshipis a part of /
  consists ofispart- of/has-part
• Nodes different concepts
• Arcs labelled links

32
Semantic Networks

• A wide variety of relationships between concepts
• To create semantic networks it is necessary to
  define a specific number of relationships and a
  coherent internal structure
• System must allow only one single method of
  description for each type of relationship
• Networks have to be subject field-specific

33
Semantic Networks

• In order to get a perfect result the end-user
  poses questions to the system
• The fragments are matched against the network
  data base
• Variable nodes in the fragments are bound to the
  value they must have in order to make the match
  perfect

34
Semantic Networks

• The success of term banks depends on several
  factors
• The semantics of the network arcs must be
  carefully defined
• System must be easy to implement and
  user-friendly
• Danger of over-complicated system that is too
  detailed

35
Compilation of TerminologyTerminological
information

• What terms are used in a terminological tool?
• The selection of the most effective terms is
  assisted by reference to terminological
  information which is collected in
  dictionaries/glossaries/term banks
• Principal factor of effectiveness type and
  quality of information

36
Terminological information

• International consensus on basic categories for
  terminological records
• entry term
• a reference number
• a subject field
• a definition
• an indication of the usage

37
Terminological information

• Customary to add indication of the sources of the
  term(s), definition, context or any foreign
  language equivalents
• It is up to the user to decide on appropriateness
  of terms

38
Corpora of raw data containing definitions,
terms, contexts
Source information
origin
type
origin type
origin type
origin type
No.
No. page
No. page
No. page
page
Conceptual Specification
Linguistic Specification
Pragmatic Specification
FL equivalent Specification
language
language
Equiv. term
language
definition
term
context
Grammatical information
Grammatical information
links to other concept
Usage note or example
synonyms
synonyms
scope notes
abbreviation
usage
abbreviation
subject field
variants
usage
variants
date type
date type
date type
date type
pool number
record number
terminologist
Housekeeping information
39
Terminological informationBasic data categories

• What information is included in a
  multifunctional term record?
• Information complex and consists of a number of
  subsets which can be compiled and processed quite
  separately.

40
Terminological informationBasic data categories

• In which categories is the term record
  structured?
• 1. source information links the term record to
  the raw data files
• 2. entry term either linguistic item or a label
  of a concept, or both
• 3. semantic and conceptual specification
  definition, a subject attribution, scope notes,
  set of links to other concepts
• 4. linguistic specification e.g. variants,
  abbreviations

41
Terminological informationBasic data categories

• 5. pragmatic specification examples of the
  context in which term occurs, usage notes
• 6. housekeeping or administrative information
  record number, name of terminologist, dates of
  first processing, up-dating of the record
• 7. foreign language equivalent specification in
  translation-orientated databases

42
Terminological informationBasic data categories

• Now let us take a closer look on the information
  categories
• Entry Term
• - most common search item
• - presented in the most relevant form
  (e.g. sing. for nouns)
• - because the distinction between concept-
  or term-orientation
• affects the treatment of
  homographs/synonyms ? decision,
• whether entry term represents concept
  or is the linguistic
• form

43
Terminological informationBasic data categories

• - In concept-orientated term banks primary
  importance on the
• definition of the concept and all terms
  matching the definition
• are grouped together ? imposes difficult
  choice of the order in which terms are listed
• - Exclusive concept orientation (e.g. NORMATERM)
  is doable
• in mono- and bilingual term banks which deal
  with subject fields of similar conceptual
  structures

44
Terminological informationBasic data categories

• - For multilingual term banks explanatory notes
  are required which indicate in every case the
  scope and degree of matching a term with the
  concept defined in another language
• - Three types of entry
• 1. simple compound or complex terms
• 2. phrases regardless of lexicalisation
• 3. sentences

45
Terminological informationBasic data categories

• Conceptual Specification
• Definition
• - first item that links entry term to the
  concept
• - can be in a style specific to the term
  bank, or extracted from
• an authoritative source
• - term banks can be classified by the way an
  entry is identified
• or explained
• - there are two major schools of thought

46
Terminological informationBasic data categories

• - The first can refer to a definition which is
  strictly limited in its validity to the range of
  texts which represent the source material for the
  term collection
• - In the second there is no restricted corpus ?
  no single valid definition in the first place

47
Terminology informationBasic data categories

• Relationships
• - most controversial and least defined category
  of information, it may indicate no more than the
  most obvious broader term
• - information could be a reference to another
  record
• Subject Field
• - terminology is divided by subject field
  before ordered in another way
• - because of the large quantities of terms it
  is advisable to introduce a classification of
  terms by subject areas

48
Terminological informationBasic data categories

• Scope Note
• - can be considered a further specification
  of subject or register
• - is intended to indicate a special field in
  application
• Linguistic specification
• Grammatical Information
• - can consist of spelling, pronunciation,
  gender for nouns,
• parts of speech (e.g. n, v, adj.),
  principal parts of verbs (e.g. infinitive, past)

49
Terminology informationBasic data categories

• Language
• - is important in term banks where it is
  combined with an
• indication of the country where it is used
• Parallel information categories to the entry
  term
• - has usually no separate record but is listed
  in an index with a
• reference to the record of the entry term
• - comprises information as spelling, expanded
  forms or
• reduced forms or synonyms
• - several overlapping categories exist
  variants, full synonyms abbreviated forms

50
Terminology informationBasic data categories

• Pragmatic specification
• ? Context
• Gives examples of the way that the entry term is
  used in a language
• Is considered a successful way of showing any
  unusual features of wordform, inflection or
  collocation
• The context should make the definition and the
  usage note complete

51
Terminology informationBasic data categories

• Usage Note
• - gives information about the way the entry
  term is used in context
• - cannot be provided in the form of examples of
  a real context
• e.g. collocational restrictions of formal
  variants
• colloquial
• slang
• mandatory
• firm-specific
• standardised
• translation
• General language dictionaries further usage
  labels as archaic,
• informal, taboo, derogatory, offensive,
  vulgar but these are
• only rarely found in terminology

52
Terminology informationBasic data categories

• Quality Label
• - term banks show in many different ways
  whether a term is standardised or and whether a
  term in a foreign language or borrowed from a
  foreign language can be considered established
  usage
• Synonyms
• terms that differ from the entry term (by usage,
  context and subject field)
• usually are fully entry terms and represent a
  crossreference in the term bank structure

53
Terminology informationBasic data categories

• Source Reference Specification
• ? Sources
• printed dictionaries rarely give an indication
  of the source
• term banks source of every relevant item of
  information is recorded
• needed for entry term, definition, context,
  translation equivalents, possibly also for
  synonyms
• Can determine the selection criteria according to
  which information is collected

54
Terminology informationBasic data categories

• Can consist of
• source origin
• ? (reliable sources in the UK
• BSI - British Standard Institute
• CEC - Commission of the European Communities
• HMS - Her Majestys Stationery Office
• ISO - International Organisation for
  Standardisation
• IEC- International Electrotechnical Commission
• - The origin of a term may be its best
  indication of quality and usage.
• - detailed reference of the source (e.g. year
  of publication ? acceptability of a term)

55
Terminology informationBasic data categories

• Source type
• - Indication of the type of document of the
  source
• ?Article in specialist literature
• ?Contracts and legal usage
• ?Governments circulars to the general public
• ?journalistic publications
• ?manuals
• ?patents
• ?publicity material
• ?research reports
• ?standards
• ?dictionary words ? should be avoided

56
Terminology informationBasic data categories

• Sources of definition and contexts
• should show different areas of usage
• Source for the foreign language equivalent
  should match the source of the entry term to make
  it suitable
• ? Source reference code or number
• large databases a separate source reference file
  that gives the full bibliographical details for
  written sources
• databases of raw data reference can be directly
  into the different/specific file

57
Terminology informationBasic data categories

• Housekeeping information (or administrative
  information)
• ? Record number
• Consists of a number for the entry, possibly with
  some subcategories
• Possible subsets of the database can be used to
  identify a topic (e.g. the terminology of a
  particular product, manual, congress or set of
  documents which can cut across subject field
  divisions)
• Such subsets are often the basis of the database
  they are isolated for separate use

58
Terminology informationBasic data categories

• ? Author of Record
• for checking the work
• author either terminologist or committee
• ? Date of Record
• date of the production of the first record
• and any up-dates

59
Terminological informationMethods of compilation

• ? Methods of compilation
• No fully acknowledged and general methodology
• Terminology compilation must become
  user-oriented!
• Serious terminology compilation is firmly
  corpus-based ? relies on the analysis of textual
  evidence
• Compilation can be a discontinous process as long
  as certain items of information which are
  connected and have an effect on each other are
  compiled at the same time
• Compilation must be seen as an ongoing revision
  and up-dating process

60
Terminological informationMethods of compilation

• Term banks softwares should provide a facility
  for prompting terminologists when building up
  terminological records.
• some form of expert system is required to control
  the work of terminologists
• If machines themselves shall be end-users of
  terminological databases there must be greater
  precision and explicitness of identification in
  the compilation of data.
• Methods to be applied in the regular compilation
  of terminology depend on
• 1. nature of data available
• 2. purpose of compilation

61
Terminological informationMethods of compilation

• Methods change with the degree of automatic
  support available ? rapid advances in the design
  of automatic tools
• no specific model is possible
• - most cases
• 1. a corpus of text is put together in
  machine-readable form (criteria
  representativeness, completeness, relevance)
• 2. corpus is fully indexed
• 3. terms are isolated and extracted
• 4. terms are sorted automatically and variously
  grouped
• 5. terms are matched with definitions

62
Terminological informationMethods of compilation

• 6. the provisional file is enlarged and
  corrected
• 7. terms are placed in relationship to other
  terms
• 8. terms are attributed to particular subject
  fields if required
• 9. a term record is created which contains
  only the term with its linguistic variants
• 10. the term record is completed with the
  addition of the house-keeping information
• - The amount and diversity of data collected
  in the term record varies according to the range
  of purposes of the data base.

63
IATE (iate.europa.eu)

• IATE ( Inter-Active Terminology for Europe)
• it is the EU inter-institutional terminology
  database system
• IATE has been used in the EU institutions and
  agencies since summer 2004 for the collection,
  distribution and shared management of EU-specific
  terminology

64
About IATE

• EU institutions and agencies involved
•          European Commission
•          Parliament
•          Council
•          Court of Justice
•          Court of Auditors
•          Economic Social Committee
•          Committee of the Regions
•          European Central Bank
•          European Investment Bank
•          Translation Centre for the Bodies of
  the EU

65
About IATE

• The project was launched in 1999 with the
  objective
• - to provide a web-based infrastructure for
  all EU terminology resources
• - enhancing the availability
• - standardisation of the information

66
About IATE

• existing terminology databases by the European
  Commission, Council, Parliament and Translation
  Centre have been imported into IATE
• ? single new, highly interactive and
  accessible interinstitutional database
• ? approximately 1.4 million multilingual
  entries

67
SDL MultiTerm 2007

• SDL MultiTerm captures, creates, manages and
  distributes terminology
• Designed for companies (e.g.marketing), which
  spend significant time and resources to create
  words which position their brand, company and
  products to the market

68
SDL MultiTerm 2007

• Concept-based terminology management
• Web- and server-based access
• Different search types (e.g. Fuzzy Search)
• Customisable data entry definitions and layouts
• It supports all worldwide languages
• Cross-references to easily link entries to each
  other (Unicode)

69
SDL MultiTerm 2007

• Benefits
• - delivers accurate and approved terminology
  with real-time verification during the
  translation process
• - quickly builds corporate glossaries
• - improves publication quality

70
Bibliography

• Sager, J. 1990. A practical course in terminology
  processing. John Benjamins B. V.
• Quah. C. K. 2006. Translation and Technology.
  Basinstoke (UK). Palgrave Macmillan.
• IATE, iate.europa.eu
• SDL MultiTerm 2007, www.sdl.com