Improving Compatibility of Terminological Collections with a Bridging Classification of Data Categories

Imp o ing compa ibili y o e minological collec ions wi h a b idging classi ica ion o da a ca eso ies IGOR KUDASHEV Uniue si y o Helsinki INTRODUCTION In spi e o he e is ence o a s anda dized se o da a ca ego ies (ISO 12620:1999) and s anda dized me hods o e minology wo k (ISO 704:2000), i is s ill ha d o ind wo e minological da abases c ea ed in wo di e en o ganiza ions which could be easily me ged wi hou loss o da a o dis o ion o he o iginal p inciples o compiia ion o each o he da abases. The mos ob ious easons o ha a e as ollows: - na ional languages and adi ions di e ; - LSP domains di e ; - backg ounds and app oachc.s o he compile s di e ; - echnical solu ions di e . A he same ime, one o he ends in he managemen o e minologi- cal esou ces and o he kinds o da a has been owa ds he agg ega ion o sca e ed esoll ces in o bigge po als o se ices, usually wi hou me ging hem physically. One o he e amples is he Eu oTe mBank po al (h p:/ / www.eu o e mbank.com) which p o ides sea ch in mul iple in e nal and ex e nal e minological da abases and can make a compila ion o he esul s. As use s'demands and olumes o he da a g ow, i becomes necessa y o p o ide ad anced sea ch which co e s all ields o he en y and no jus headwo ds and o ailo en ies acco ding o use s'p e e ences. In his a icle, we p opose a classi ica ion o da a ca ego ies which can se e as a b idge be ween e minological collec ions. The classi ica ion add esses he ollowing p oblems ela ed o he managemen o e minoiogy collec ions wi h di e en se s o da a ca ego ies: - agg ega ion and me ging o e minological esou ces; - o ganiza ion o sea ch which co e s all ields o he en ies; - ailo ing he en ies om mul iple collec ions in acco dance wi h use s' o e e ences. 36 Igo Kudashe Imp ouing Compa ihili y o 'le minological Te minological cla abase may con ain di e en ypes o da a in addi ion o e n inoiogical da a p ope , such as in o ma ion abou sou ces. use s and e minology managemen ansac ions. In his a icle we ocus on he clas- si ica ion o da a ca ego ies ela ed o he desc ip ion o LSP exp essions. WHAT IS A DATA CATEGORY? Di e en ypes o da a a e placed in di e en ields in e minological da abases. Da a ca ego y is he esul o speci ica ion o a gi en da a ield (lSO 1087-2:2000: 13). Te ninologis s use di e en me apho s in o de o explain he concep o a da a ca ego y. One me apho is ha o a wa d obe , he e di e en d awe s a e usecl o s o e di e en ypes o cio hes. Ano he me apho is a shopping baske he e eacl- kind o goods is w apped in i s own package so ha hey do no ge mixed. TYPICT L A4ISN4ATCIIES BET VEEN DATA CATEGORIES Da a ca ego ies a e he esul o da a classi ica ion. Da a can be classi ied in many di {e en ways depending on he iews o he classi ie and he needs o he use s. Using he u'a d obe me apho one can say ha cli e en wa d obes come u'i h di e en nl mbe o d awe s which may be o di e - en shape. size, e c. A leas si ypes o misma ching a e possible be ween cla a ca ego ies in e minological da abases: - misma ch o he names o da a ca ego ies; - misma ch o he "sizes" o da a ca ego ies, i.e. di e en "g anuia- ion": - mis a ch o he "places" o da a ca ego ies. i.e. hei loca ion in he classi ica ion scheme. : - misma ch o he classi ica ion p inciples (o e lapping); - nisma ch o he con en s o da a ca eso ies; - mixed L ases. Below a e a ew examnles which illus a e some common misma ches. Case 1: da a ca ego ies a e named di e en ly. Example: da a ca ego y is called no e in one da abase, commen in ano he , ema k in he hi d one and NB in he ou h one. Te minologija | 2009 I 16 37 Case 2: he name o a da a ca ego y is " sed in di e en senses. Example: da a ca ego y synonym ma co espond o " ull synonym", "nea -synonym" o " ull o nea -synonym'i Case 3'. names o da z ca ego ies a e " alse iends'i Example: English sbb euia ion and ac onym as cle ined in he ISO s anda d (ISO 12620:1999:6-7) and Russian a66peauamypa and oKpoHuM co espond o each o he c osswise (Kudashe & Haju in 2003:104-105). Case 4'. g anula ion o da a ca ego ies is di e en . Example: da a ca ego y abb euia ed o m o e m is spli in o i e sub- classes (abb euia ion. sho o m o he e m, ini ialism, ac onym and clipped e m) in ISO 12620:1999, bu he e is no such di ision in ISO 12616:2002 (" T ansla ion-o ien ed e minog aphy"). Case 5: o e lapping o da a ca ego ies. Example: da a ca ego es example and con ex o e lap. Some e amples a e con ex s and some con ex s may se e as exan ples, bu he wo ca ego ies a e no iden ical. Case 6'. he same da a ca ego y is pu unde di e en b oade ca ego ies. Example: da a ca ego y con ex is conside ed concep - ela ed da a in ISO 12620:1999, appa en ly because con ex s may p o ide addi ional in o ma- ion on he concep . Howe e , a mo e conlmon nc ion o con ex s is o p o ide in o ma ion abou e m usage and colloca ions, so "many da abases classi y con ex as a e n- ela ed ca ego y" (ISO 12620:1999:25). Case 7: language, sign sys e n o no a ion o he da a di e . Exa nple'. a pa o speech may be coded as noun in one da abase, n. in ano he , subs/. in he hi d one and as a g aphical symbol in he o h one. Case 8: iden ical o simila alues o da a ca eso ies a e used in di e en meanings and di e en connec ions. Example: alue neologism in he ISO 12620:1999 e m p ouenance d.a a ca ego y sounds like a ch onological label n'hile in ac i e e s o he me hodology e.mployed in c ea ing he e m. 38 Igo Kudasl e I Imp o 'i g (,'on po ibili y o Te minol >gical.. Case 9: alues o semi-closed classes di e due o di e en classi ica ion o linguis ic phenomena in di e en lang ages and di e en adi ions. Example: in ISO 12620:1999 labels o desc ibing LSP e p essions be- longing o he "lowe s yie" include slang egis e and uulga egis e . They oughly co espond o labels npoQ ecc uoua naua i )Kap zoH L3 M (p o essional slang) and npoc!eccuoHanbHoe npocmopeque (p o essional colloquialism) in Russian. In Finnish, howe e , he e is only one ca ego y, amma islangi (p o essional slang; see Sanas o yon k:isiki ja 1,989: 12). I should also be no ed ha some da a nay be exp essed implici ly in e minological collec ions. Example l: he p esence o he label obsole e in some cases and i s in o he s in a e minological collec ion implies ha e ms which ma ked wi h he label. belong o he ac i e LSP s ock. Example 2: i wo o mo e e ms a e pu in he same en y, his implies ha hey a e synonyms o equi alen s. absence a e no usually Da a exchange o agg ega ion o e minoiogical esou ces n ay equi e ha such implici da a is made explici . Fo example, i da a is s o ed in he o m o on ology a he han s a ic en ies, hen implici in o ma ion abou synonyms and equi ,alen s men ioned abo e has o be made explici and sa ed du ing he "dissembling" o he en ies. ISO INVENTORIES AND CLASSIFICATIONS OF DATA C]ATEGORIES ISO 12620:1999 speci ies a se o da a ca ego ies o eco ding e mino- logical in o ma ion. I does no p esc ibe wha da a ca ego ies should be used bu a he se es as an in en o y. This se o da a ca ego ies is being b oadened in he ISOca p ojec ( w'w.isoca .o g) which documen s widely accep ed linguis ic da a ca ego ies. While he ISOca p ojec is undoub edly use ul o linguis s who ge a chance o be e ca ego ize and de ine linguis ic concep s and o designe s o e m banks who can pick eady da a ca ego ies om a as in en o y. i is unlikeiy o b ing elie o he p oblem o da a exchange, agg ega ion and ull-en y sea ch in mul iple da abases. Fi s ly, i can' p e en compile s o e minological p oduc s om Te minologija | 2009 | 16 39 classi ying da a in a di e en way and om using hei o ,l,n da a ca ego ies. Secondly, he mo e da a ca ego ies a e used in e minological collec ions, he no e di e se hey beco ne, making da a exchange e en mo e chailenging. The lis o da a ca ego ies is in ini e in heo y and qui e as i p ac ice. Fo example, he lis o da a ca ego ies eia ecl o e minology in he lSOca/ in en o y al eady exceeds 500 i en s. one o he p ac ices used in he ISoca p ojec can inc ease his numbe nany imes. We e e o p esen ing alues o clo.sed and semi-closed classes as indi .idual da a ca ego les. Fo example, alues o he da a ca ego y egis e , s ich as uulga Regis e . s/cngReg s/e . e c. a e now p esen ed as da a ca ego ies in liei own igh . Likewise, da a ca ego y eliabili yCode is spli in o eliabili yCoclel, elia- bili yCode2, e c. The e ca n be di e en opinions whe he his is easonable om he p ac ical poin o ie , bl da a can be o ganized in many di e e 'ays, so he e a e no o mal es ic ions o spli ing da a ca ego ies ail |e way o he p imi i e classes which can only accep alues 1, :s and no. I is clea . howe e , ha his ealm o da a ca ego ies needs p ope s uc u ing and classi ica ion in o de o emain nanageable and lyell-o ganizeC. Meanwhile. classi ica ion o da a ca ego ies p oposed in ISC i2520:1g99 is p oble na ic in se e al espec s. To s a wi h, he e a e -somc incolsis en- cies conce ning he p incipal di ision o he da a. Acco cling io sec ion 6.2 (Ty'polog1, o da a ca ego ies). da a ca ego ies a e di icied in c h ee najo g oups: e m and e n- ela ed in o ma ion. clesc ip i e Ca a 3nd adn inis- a i e da a. Howe e . in Annex D (S s ema ic lis ing o da. o ca ego - es) he second g oup is called Da a ca ego ies ela ed o concep iesc iS: ion. A |e sa ne in e, his g oup cou ains s bg clup Nole whicli "s ands a.ioi e ecanse i can be associa ed wi h any one o he o he ca ego ies and lhe e o e can- no be subo dina ed o any o he speci ic subg oup" (ISo i?.0?,a: 999:4). I we assun e l- a he in ended di ision included ou e ,:ups; ie n ancl e m- ela ed in o ma ion, concep - ela ed in o na ion, adminis a iye cl; a and No/e, his classi ica ion s ill aises many ques ions. Tc. naii:e a {e.l ; - why examples and con ex .s a e concep - ela ecl da a and no e m- ela ed da a? c . he desc ip ion o he con ex da a ca ego y:,'A ex o pa o a e in which a e n occu s" (ISo 12620:l 999: 25). - A hy s nonym and equiualence a e e n- ela ed da .: ,uhilc o i e - wise e e y hing ela ed o he meaning is concep - e e ec- da a? c . desc ip ion o he da a ca ego y deg ee o equiualence: "The ex en o which he in ensions o wo o nlo e concep s i-."",e iap" (lso 72620:1999:21). '+0 Igo Kudi she I Imp ouing (.)ompa ibii y o. Te minological - hy an onym and homog aph a e ad ninis a i e da a and no e m- o concep - ela ed da a? - Why audio, uideo, e c. a e (j - s ) concep - ela ed da a? - Wha is he exac cle ini ion o adminis a i e da a? Wh his class includes such he e ogeneoLls ca ego ies? Di ision o e minological da a in o concep - ela ed ancl e m- ela ed may be use ul om he echnical poin o ie u because i suppo s he concep -o ien ed app oach which ednces he numbe o ela ions be ween e ms by linking synon lnous enns o he sa n€. concep . How-e e , such di ision in gene al and i s implemen a ion in ISO 12620:1,999 in pa icula ma p esen a challenge o common use s o e minological da abases - ansla o s ai d domain e pe s. Indeed, i is no easy o comp ehencl why synon),ms should be sea ched in e m- ela ed da a, b an onyms in adminis a i e da a and examples in concep - ela ed da a. Common use s wo k wi h e ms - wo ds and wo d combina ions, so o hem i is o e na u al o speak abou he meaning o a e m, synonyms o a e m, exampies o e m usage. e c. This i npiies i a classi ica ion o da a ca ego ies aimed a p o iding ll-en y sea ch in mul iple collec ions should be e m-o ien ed, in ui i ely clea and based on common linguis ic ca ego ies. RRIDGING MISMATCHES BETWEEN DATA CATEGORIES The p oblem o misma ches be ween da a ca ego ies can be sol ed wi h he help o mapping. I n isma ches a e nominal, o example, i names o da a ca ego ies di e o he same con en s a e p esen ed di e en ly, di ec mapping be ween da a ca ego ies can be used. Once he di ec mapping is applied, he da a in he agg ega ed esou ce can be sea ched wi h he same p ecision as in he o iginal da abases. Mo e subs an ial di e ences be ween da a ca ego ies equi e inding a common denomina o h ough he napping be ween classi ica ions o da a ca ego y se s. The use o a common denomina o somewha dec eases he p ecision o he sea ch, bu his is he only way o p o ide co n non sea ch o e minological eso l ces wi h di e en s uc u e. I he use is no eady o sac i ice he p ecision o he sea ch, he has o pe o m sepa a e sea ch in each indi idual collec ion. As wi h da a e change in gene al. sing an in e media e o ma , i.e. some b idging classi ica ion o da a ca ego ies, is mo e e ec i e in he long 41 Te minologija | 2009 I 16 un han mul iple mappings be ween di e en classi ica ions. The gene al p inciples o he classi ica ion o da a ca ego ies in ended o se ing as a common in e ace be ween o he classi ica ions a e as ollows: 1. Since classes o he classi ica ion a e supposed o se e as comlnon denomina o s. hey ha e o be on a highe le el o abs ac ion han mos "p imi i e" da a ca ego ie's which a e no subdi ided u he . A he same ime he le el o abs ac ion may no be oo high because he use s would no be in e es ed in a classi ica ion which is oo gene al. In p ac ice a clas- si ica ion wi h wo le els o abs ac ion is su icien . 2. The classi ica ion sho ld co e all ypes o LSP exp essions which a e ypically desc ibed in e minological cla abases, including e m elemen s, p ope names, nomencla u e, se ph ases, e c. 3. The classi ica ion should con ain only hose da a ca ego ies which a e di ec l ela ed o he clesc ip ion o LSP exp essions. Desc ip ion o echnical and ad ninis a i e aspec s (e.g. n edium, encodings, sou ces. eli- abili y, e c.) is a di e en ask. 4. The classi ica ion should be hie a chical, bu se e al bases o di ision should be allowed on he same le el o abs ac ion. 5. The ciassi ica ion should be ex ensible. i.e. i should con ain ca eqo "o he " on each le el o he hie a chy. These p inciples ha e o be combined wi h he equi emen o use - iendliness men ioned abo e. In ou opinion classi ica ion based on lin- guis ic unc ions o he da a is a pa icula ly s ong candida e in he la e espec . In he b idging classi ica ion basecl on he linguis ic unc ions he ocus is shi ed om he names o da a ca ego ies o he unc ion o unc ions o he da a hey con ain. A compa ison which can be made he e is a lib a y wi h i s alphabe ic and subiec di ec o ies. Sea ch by he names o da a ca ego ies is simila o e sea ch wi h he help o an alphabe ical di ec o y. Thc' i le gi es so ne clue abou he con- en s o he book, bu some imes i can be misleading. Besides, use s canno guess all he possible i les which co e a ce ain opic. Alphabe ic di ec- o y is use ul in cases when he use s knol o su e wha i em(s) hey wan o loca e. Mos lib a L se s, howe e . s a hei sea ch wi h he sub ec di ec o y because hey do no know be o e'hand wha i ems co e he opic hey a e in e es ed in. Te minological po al which agg ega es mul iple e n- inological collec- .1 -) Igo Kudashe Imp ouing Con pa ibili y o Te minological.. ions l i h di e en se s o da a ca ego ies poses he same p oblem o use s as he ealm o books on he shel es o a lib a l'. Use s know wha kind o in o ma ion hey a e in e es ed in (g amma . meaning, L sage, e c.), b hey do no necessa ily know in wha da a ca ego ies his in o ma ion can be ound. In di e en e minological collec ions simiia ypes o da a ma eside in di e en da a ields. Fo exa nple, in o ma ion abou he a eal s a us o an LSP exp ession can be ound in such iields as usage, egional label, language symbol, e c. I he sea ch is based on he linguis ic unc ions o he da a. use s don' need o ca e abo he e ac names o da a ca ego ies. They jus speci y ha hey a e looking o in o ma ion abou a eal s a us, meaning o syno- ny ns, and he e ninology managemen sys em loca es and displays ull o ab idged en ies in di e en da abases which con ain he speci ied ype o in o ma ion. Sea ch by he unc ion may and should be complemen ed by sea ch by he name o he ields o hose l se s who kno ,l exac ly in wha ields hey wan o sea ch. Da a ields, like lib a y i ems, may con ain di e en kinds o linguis ic in o ma ion. Fo ins ance, example is a ypical mul i unc ional ield which may con ain in o ma ion abou o m, meaning and usage o a e m, in di e en p opo ions. Each cla a ca ego y may be desc ibed as ha ing one o se e al iunc ions. I should also be possible o speci y he deg ee o hich a pa icula da a ca ego e lec s ce ain unc ions. This cha ac- e is ic may be e bal (e.g. p ima y and seconda y unc ion) o nl me ical (e.g. 0-100%). The easies way o do he napping o a pa icula cla a ca ego y se o he in e media e classi ica ion is o conside he con en s o da a ields o a pa icula ype uni o m and make a simple able o co espondence. A mo e p ecise bu also mo e co nplica ed mapping would allow he compile s o speci y cle ia ions o indi idual da a ields om he al es used in he' global able o co espondence. Fo example, con ex s usually p o ide in o ma ion abou e m usage and meaning. I is easonable o include hese ypes o in o ma ion in he global able o co espondence as hey pe ain o e e y con L'x ield. Hol ,'e e , some con ex s may also con ain encyclopedic in o ma ion. This occasional use o encyclopedic in o ma ion may be ma kecl locall- . o he ieid le el. 43Te minologija | 2009 16 CLASSIFICATION OF DATA CATEGORIES BASED ON LINC}UISTIC FUNCTIONS OF THE DATA Since LSP exp essions a e linguis ic signs, in o ma ion abou hem can be di ided in o in o ma ion abou hei o m, meaning, L sage, ela ions o o he uni s. o igin and de elopmen . Below each o hese classes is de- sc ibed in de ail and some examples o he da a belonging o hem a e. p o- ided. Please see Appendix 1 o he compac e sion o he classi ica ion. Da a ela ed o o n LSP exp essions ha e wo o ms - w i en and o al. Besides, da a ela ed o he o m o an LSP exp ession may be subdi ided in o h ee classes: - da a ela ed o he canonical o n; - da a ela ed o he o ma ion o he uni ; - da a ela ed o he in lec ion o he uni . DATA RELATED TO CANONICAL FOR]Vl Canonical o m is he o m in which he headwo d is gi en in he da- abase. I se es as " ep esen a i e" o o he o ms. Fo example, in mos Eu opean li nguages he canonical o m o nouns is nomina i e singula and o e bs in ini i e p esence. Howe e , ules o choosing canonical o m a y in di e en language's and lexicog aphical acli ions. He e a e a ew exa nples o he da a ela ed o w i en canonical o m: - ype o exp ession by i s w i en o m (e.g. ull o n, abb e ia ed o m, sy nbol, o mula. e c.); - spelling o he o m: - spelling a ian s o he o m; - hyphena ion. Examples o da a ela ed o o al canonical o m: - ype o exp ession by i s o al o m (e.g. ini ialism, ac onym); - p onuncia ion; - p onllncia ion a ian s o he o m; - syllabi ica ion. As one can see. some da a ca ego ies may ela e o bo h w i en and o al o ms. Fo example, indica ing ha a pa icula exp ession is an ini ialism, i.e. an abb c. ia ed o m made o ini ial le e s o he ull e m, in which hese le e s a e p onounced indi idually (e.g. Uni ed Na ions - UN), p o ides in o ma ion abou bo h w i en and o al o ms. Besides, i may Igo K - dashe I Imp ouing Compa ibili ' o Te minologicul.. 4.1.8. Da a ela ed o egis e es ic ions. < e m(Jsage: es ic ions.. egisle ) 4.7.9. Da a ela ed o p o essional g oup es ic io s. < e mlJsage: es ic- ion s : p o e s si onal G ou p> 4.1.10. Da a ela ed o combina o y es ic ions . l e m(Jsage: es ic ions: combina o y) 4.1.11. Da a ela ed o compliance es ic ions. e m -Jsage: es ic ions: compliance> 4.2. Da a ela ed o equency o use. l e mLlsage: equenc > 5. Da a ela ed o sys ema ic ela ions. < e mRelo ions> 5.1. Da a ela ed o synonymous ela ions. < e mRela ions: synonymousRela ion) 5.2. Da a ela ed o an onymous ela ions. l e mRela ions: an onymousRela ion> 5.3. Da a ela ed o homonymous ela ions. < e mllela ions: homonymousRela ion> 5.'1. Da a ela ed o pa onymous ela io s. < e mRela ions: pa onymousRela ion) 5.5. Da a ela ed o gene ic ela ions. < e mRela ions: gene ic.Rela ion) 5.6. Da a ela ed o pa i i e ela ions. < e mRela ions: q i iaeRele ion> 5.7. Da a ela ed o non-hie a chical on ological ela ions. < e mRela ions: non- hi e a c hic alO n ologic alR el q ion> 5.8. Da a ela ed o equi alence eia ions. < e mRela ions: equiualenceRela ion> 6. Da a ela ed o o igin and de elop nen . ( e no iginAndDeuel- opmen ) 7. o he ypes o da a ela ed o he desc ip ion o an LSp exp es- sion < e m O he Rela e dD a a > APPENDIX 2. EXAMPLE OF MAPPING OF SOME ISO 12620 1999 DATA CATEGORIES INTO THE I,UNCTIONAL CLASSIFICATION OF TERMINOLOC;ICAL I)ATA See Appendix 1 o he dec yp ion o he o mal ep esen a ion o he ca ego ies which a e used o he sake o b e i y. eues ion ma k a e a da a ca ego y means ha he p esence o he speci ied kind o da a is subjec o he in e p e a ion o he da a ca ego y. I subca ego ies can be mapped in he exac same way as l- ei pa en ca ego y, only he pa en ca ego y is desc ibecl. Te minologija 12009 | 16 51 D e o space li ni a ions only he i s subg oup o cla a ca ego ies om he ISO I2( 20:1999 s anda d is co e ed. A. I e n No e: e m (headu..o d) lies ou side hc- scope o classi ica ion which co e s da a ca ego ies ela ed o he desc ip ion o LSp exp essions. Howe e , echnicall- da a ca ego y e m is iclen ical o he w i en o m o he e m. Func ional classi ica ion: no applicable o < e nFo m: w i enFo m> 4.2.I e n ype No e'.in he ISo 72620:1999 s anda d his ca ego y includes a ious kinds o LSP e p essions and may con ain di e en ypes o da a. see subca e_ go ies o mo e in o ma ion. A.2.l.l main en y e n No e: in o ma ion abou he s uc u e o he en y is no da a ela c.d o he desc ip ion o LSP e p essions. Howe e , his ca ego y indi ec ly e lec s p e e en 'e. Func ional classi ica ion: no applicable o < e mLlsage: es ic ions: compli_ 0nce> 4.2.1.2 synonym No e: when opposed o main e.n e m, his ca ego e lec s p e e ence. I s p ima y nc ion is, howe e' , e lec ion o s nonymous ela ions. Func ional classi ica ion: l e mRela ions: synonymousRela ion>; ( e mlJsage: es ic ions: complianc e> | 4.2.1 . 3 quas i-synon). n No/e: san e as p e ious. Func ion,l classi ica ion: < e mRela ions: s nonymousRela ion>; < e mI)sage: es ic ions: con pliance>? 4.2.1,4 in e na ional scien i ic e n No e: depending on he in e p e a ion o his ca ego y and i s co' en s. his da a ca ego y n ay p o'ide in o ma ion abou co np iance, la'guage es ic ions. o igina ion and de elopmen . Func ional classi. ica ion: < e mLlsage: es ic ions: compliance); < e m(Jsage: es ic ion s : lan guage> ? i < e m () iginAn d D eu elop m en > | 4.2.1 .5 co nmon name No e: common name is a sy'onyn o an in e na ional scien i ic c. m ha is 'sed in gene al cliscou se. This da a ca ego y nay p o ide in o ma ion abou es ic ions ela ed o egis e , compliance and p o essionai g oup. 52 Igo Kudasl e I In p ouing Compa ibili l, o Te ninological Func ional classi ica ion'. < e mUsage: es ic ions: egis e > ; < e mUsage: e- s iclions: compliance> : < e musage: es ic ions: p o essionalG oup>? A.2.I.6 in e na ionalis n No e: depending on he con en s his da a ca ego n a p o ide in o ma io - ela ed o language es ic ions, e nl o ma ion, o igina ion and de elop- men . Func ional classi ica ion: < e mUsage: es ic ions: Ianguage>i < e mFo m: e n F o m a i o n > l', < e m() i gi n An l D eu e I o p m e n > ? A.2,L.7 ull o n No/e: his da a ca ego y p o ides in o ma ion abou he ype o e m by i s o m. Fu nc ional c Ia s si ic a ion'. < e mFo m : e m Ty p e'> A.2.1.8 abb e ia ed o n o e m No/e:depending on he con en s, his da a ca ego y and all i s subca ego ies (A.2.1.8.1 abb e ia ion. A.2.1.8.2 sho o m o e m, A.2.1.8.3 ini- ialis n, A.2.1.8.4 ac ony n and A.2.1.8.5 clipped e n) may p o ide in o ma ion ela ed o ype o e m by i s o m, o al o m: e m o ma ion, o igin and de elopmen . Func ional classi ica ion: < e mFo m: e mType> ; < e mFo m: o alFo m>? ; <. e m Fo m: e mFo ma i on>'/'. <, e m() iginAnd D e elopm en > A.2.I.9 a ian No e: depending on he con en s, his da a ca ego y may p o ide in o ma- ion abou p e e ence, e m o ma ion, o igin ancl de eiopmen . Func ional classi ica ion: < e mUsage: es ic ions: compliance>; < e mFo m: e mFo ma io n> ? ; < e mO iginAn l D eu elopmen > ? A.2. 1. 10-A .2.1.I4 ( ansli e a ed o n, ansc ibed o n, o nan- ized o m, sy nbol and o mula) No/e: hese da a ca ego ies p o ide in o na ion abou he ype o e m b,,,- i s o m. Func ional classi ica ion: l e mFo m: e mType) A. 2. 1. I 5-A .2.L.I7 (equa ion, logic al exp ession, na e i als nanage - men ca ego ies) No e: hese uni s can no be headwo ds in e minolosical clz abases. Func ional classi ica ion: no applicable. A.2.1. 18 ph aseological uni No e:in ISO 12620:1999 his ca ego y is spli in o h ee s bca ego ies:col- loca ion, se ph ase and, synonymous ph ase. Colloca ion ca no be coun ed le minologija I 2009 I 16 53 as a ph aseological uni as i does no co espond o he de ini ion o a ph aseological uni p o ided in he s anda d. synonymous ph ase seems o be an unl ecessaly ca c.go li The same in o ma ion can be exp essed wi h wo o he da a ca ego ies: se/ ph ase and synonym. Func ionsl classi 'ica ion: no applicable. A.2. 1. 18. colloca ion No e: his da a ca ego y p o ides in o ma ion abou combina o y es ic- ions. Func ional classi ica ion: 1 e m[Jsage: es ic ions: combina o y> A.2.1.I8.2 se ph ase No e; his cla a ca ego y p o ides in o ma ion abou he ype o e m by co espondence o i s mc-aning o i s o m. Fu c ional classi ica ion'. l e mMeaning: e nType> A.2.18.3 synony nous Ph ase No/e: as was s a ed abo e, his ca ego y is p obably unnecessa , . I used. i p o ides in o ma ion abou he pe o e m by co espondence o i s meaning o i s o m and aiso abou synonymous ela ions. F nc ionc l classi ica ion'. < e mMeaning: e mType>; < e mRela ions: synonymousRela ion> A.2.1.19 s anda d ex No e: s alda cl e s can no be headwo ds in e minological da abases. Func ional classiiica io ?: no applicable. TERMINIJOS RINKINIlJ SUDERINAMUMO GERINIMAS TAIKANT DUOMENll KAIEGORIJ l KIASITIKA(IJ4 Vie na ii in o l acijos i pac: e minologijos iS ekliq pld os endenciiq y a a ski q duomenq baziq ju . gi nas ididclius po al , s i sqsaiq a p okiq po z lq k imas. To pa yz l s gali bu i Eu >Te mBankas (h p:,/i w$ ou o e mbank.com). Dideian in- o maci jos apin iiai 1 a o oiq okie ls p oduk an s kelian ien s elkala imams, auga po eikis lZ ik in i iilpis ing paieik4, api nandi4 isus iodyn l s aipsniq laukus. i gali- - ybg pa eik i ik as in o nacijos ka cgo ijas, ku ios uo me u clomina a o oi4. Siems uZda iniams sp qs i eikalinga duomen l ka ego ii l klasi ikaciia, ku i4 bu q galima 2ik i kaip e n. inologiniq baziq, u indiq ski ingq s uk .1 4. ,,b.-n.1 q I'a dikli'l S a bus okios kla si ikacijos eikala .imas jos pap as mas i aiSkun as paP as ienls ninolo- gijos p odL k l -a o ojan s. S aips . yle ap iana ling is ine nis duomenq unkciio nis pa em a klasi ikaciia gal q ap i okiu ..bend u a dikiiu'l 51 Igo K clashe | ln p ouing Compu ibili y' o 'le minological"' ncn0n b30BAH 14 E MTTAK'lA(C0B AAH H blX KAK ( PEACTB0 yny' il EH 148 C () B M E C T 14 M O C T 14 T E P M 14 H OII O T 14 II E C K 14 X K ON T E K U }.I R O.qnoi il 3 TeH,4eH 1x i pa:nul z uHQopua luoHub x pec pcoB B le oM - ' epM lHo- / o u iecKux B qac noclu c a; o o6 e;lu cH 4e'pa3po3HCHHh x 6a: asHmx R (py Hb € op a bl co3AaHI4e o6l1ux uHlep<|eiicon K HuM. Ilpu ' epo ' Moxe cn xcu ' Enpo- TepuEaux (h p://ww ' '.eu o e mbank.com). lo , epe pocla o6ne-uou lu lop ' aqu l pe6o aHuii lo. b3oBa e eii nogo6H , M npoilyK' aM pac e no pe6Hoc s o6ccne-- qeH u pacil upeHHo o no , cKa, oxBaTb Ba o[Ic o Bce ol { c onaP , x cll ei, .1 TaKxc Bo3Mo)KHOc u o o6paxeu: , n 4l h lex (a e opl.{ii ( op u aql [! I{olop ] e n 4aHn , ii MoMeHT 14HTepecyio nonbioBa e,' . 3 u :a. 1a- l pc6 ' u ul; . a n l xaaccuc],l xaqll nu lop.uaqno Hb x Ka e opl ii. xo p. l , o ; a 6 , c/ yl< l 'l, <o6 q ' nl 3Ha uena c,' eMD .q q epMuHo,' o , qecKux 6a: c pa:nuuHoii clpy poii. He- anona)K { , M pe6oBaH ' eM nogo6 oi <nacc . c uxaqun , B .se c cc poclo' a 14 o ]oc h x npoc blx oJ - sona e eii ' 'epM . Ho o . ecKI4x lpoAyKToB. B c a ne o[ cb Bae c o 'Il H . 3 BO:lMo)K- HblX KaHA . AaTOB Ha pOni' uO6 qe o 3HaMe a' 'elao - xnaccn{ 4Ka l 4- l. ocHoBaHHa a J l H R {c { ec ux < yH l . x .{aHHb . Gau a 2009-10-i-l Igo Kudashe Uni c si o Helsinki Palmenia Cen e o Co i uing EdL ca ion P.O. Box 239 (Pa aa iken d 6) FIN-45 100 Kou ola. Finland E-mail: [email p o ec ed] Te minologiia I 2009 | 16 55