Conceptual Structuring through Term Variations
B´eatrice Daille
IRIN
University of Nantes
France
daille@irin.univ-nantes.fr
Abstract
Term extraction systems are now an inte-
gral part of the compiling of specialized
dictionaries and updating of term banks.
In this paper, we present a term detection
approach that discovers, structures, and
infers conceptual relationships between
terms for French. Conceptual relation-
ships are deduced from specific types of
term variations, morphological and syn-
tagmatic, and are expressed through lexi-
cal functions. The linguistic precision of
the conceptual structuring through mor-
phological variations is of 95 %.
1 Introduction
Term extraction systems are now an integral part
of the compiling of specialized dictionaries and up-
dating of term banks. Several tools exist either for
extracting or structuring terminology (see (Cabr´e
et al., 2001) for a review of the current systems).
Systems for identifying conceptual relationships are
generally based on external evidence: (Cort`es and
Cabr´e, 2002) define a catalogue of linguistic mark-
ers to detect conceptual relationship such as simi-
larity or inclusion relationships. Similarity is de-
tected by the prototype linguistic expression: to be
similar to. Other systems relies on internal evi-
dence. (Morin and Jacquemin, 1999; Hamon and
Nazarenko, 2001) structure complex terms (multi-
word terms) with the help of lexical databases or
general dictionaries. The relationships handled are
limited to synonymy and hyperonymy. (Grabar and
Zweigenbaum, 2000) identify morphological fami-
lies of word forms applied on a medical thesaurus
with a precision of 92 % such as (arthrite “arthri-
tis”, arthrose “arthrosis”, arthropathie “arthropa-
thy”, a0a1a0a1a0 ). They also make deductions on their con-
ceptual relationship with other lexical units and their
contribution to the overall knowledge organization
of a specialized field, namely medicine. The con-
ceptual relationships identified are synonymy, refer-
ence and hyperonymy.
In this paper, we present a term detection approach
that discovers, structures, and infers conceptual rela-
tionships between terms for French. Conceptual re-
lationships are deduced from specific types of term
variations, morphological and syntagmatic, and are
expressed in terms of lexical functions (Wanner,
1996). Term variations have already proved their re-
liability in information retrieval: (Jacquemin, 2001)
considers different types of terminological variants
including syntactic variations to perform accurate
text indexing. In the remaining sections, we present
some conceptual systems, provide a linguistic typol-
ogy of term variations and describe the two steps of
our approach. In the final part, we give results and
briefly discuss them.
2 Conceptual systems
Terms are generally classified using partitive and
generic relationships to be presented in a thesaural
structure. But other relationships exist, the so-called
complex relationships (Sager, 1990, pages 34-35)
which are domain and application dependent. Ex-
amples of such complex relationships are:
FALLOUT is caused by NUCLEAR EXPLOSION
COAL-MINE is a place for COAL-MINING
Studying term formation, (Kageura, 2002) intro-
duces intra-term relationships dealing with complex
terms and defining the role of the determinant with
respect to the head noun. In computer program, pro-
gram is the head noun and computer the determi-
nant. As program is intended for computer, a “des-
tination” intra-relationship occurs between program
and computer. (L’Homme, 2002) used lexical func-
tions to represent various types of relationships via
an unique formalism for a computerized version of a
dictionary. These relationships are of several types:
a0 paradigmatic such as the generic (Gener) or
antonymy (Anti) relationships:
Gener(retail sale) = sale,
Anti(retail sale) = wholesaling;
a0 derivational such as nominalizations (S
a1 , Va1 ):
Sa1 (to program) = programmer,
Va1 (to program) = programming;
a0 syntagmatic (Bon, Real
a2 ):
Bon(software) = performing,
Real a2 (programme) = to run [DETa3 ].
These conceptual relationships are assigned manu-
ally to terms or sets of terms. We propose to au-
tomatically assign conceptual relationships to com-
plex terms through their variations.
3 Concept identification through Term
variations
For complex terms identification, it is necessary
to first define syntactic structures which are poten-
tially lexicalisable. These complex sequences are
so-called “base-terms”.
3.1 Base terms and their linguistic Variations
For French, the syntactic structures or patterns of
base-terms are:
Noun1 Adj emballage biod´egradable (biodegrad-
able package)
Noun1 (Prep (Det)) Noun2 ions calcium (calcium
ion) prot´eine de poissons (fish protein), chimio-
prophylaxie au rifampine (rifampicin chemo-
prophylaxis)
Noun1 `a Vinf viandes `a griller (grill meat)
These base structures are not frozen structures and
accept variations. Terminological variation in texts
is now a well-known phenomenon estimated from
15 % to 35 %, depending on the domain reflecting by
the texts and the different kinds of variants handled.
For acquisition, it is essential to identify extensively
all the concepts represented by terms in textual data.
Thus, only term variants which can preserve the
base-term semantics and thus refer to the same con-
cept are taken into account in a first step. Two se-
quences such as histamine pr´esente dans le vin (his-
tamine which is present in wine) et histamine du vin
(histamine of the wine) refer to the same term his-
tamine du vin (wine histamine); but, the sequences
produit `a surgeler (product to be frozen) and pro-
duit surgel´e (frozen product) refer to two different
terms linked by an aspectual relationship.
We present now a linguistic typology of base-term
variations for French:
Graphical case differences and presence of a op-
tional hyphen inside the Noun1 Noun2 structure.
Inflexional orthographic variants gathering to-
gether inflexional variants that are predictable such
as conservations de produit (product preservations)
or unpredictable such as conservation de produits
(product preservation).
Shallow syntactic The shallow syntactic varia-
tions modify the function words of the base-terms.
There are three kinds of internal syntactic variations:
is-1 variations of the preposition: chromatographie
en colonne (column chromatography) a4 chro-
matographie sur colonne (chromatography on
column);
is-2 optional character of the preposition and of the
article: fixation azote (nitrogen fixation) a5 fix-
ation d’azote (fixation of nitrogen) a6 fixation
de l’azote (fixation of the nitrogen);
is-3 predicative variants: the predicative role of the
adjective: pectine m´ethyl´ee (methylate pectin)
a6 ces pectines sont m´ethyl´ees (these pectins
are methylated).
Syntactic The shallow syntactic variations modify
the internal structure of the base-terms:
S-1 Internal modification variants: insertion inside
the base-term structure of
a0 a modifier such as the adjective inside the
Noun1 Prep Noun2 structure: lait de bre-
bis (goat’s milk), lait cru de brebis (milk
straight from the goat);
a0 a nominal specifier inside the Noun Adj.
These specifiers belongs to a closed list
of nouns such as type, origine, couleur
(colour): prot´eine v´eg´etale “vegetable
protein” a6 prot´eine d’origine v´eg´etale
“protein of vegetable origin”.
S-2 Coordinational variants: head or expansion co-
ordination of base term structures and enumer-
ation:
analyse de particules “particule analysis” a6
analyse et le tri de particules “particle sort and
analysis”
alimentation humaine “human feeding” a6 ali-
mentation animale et humaine “human and an-
imal feeding”.
Morphosyntactic The Morphosyntactic varia-
tions modify the internal structure of the base-terms
and its components are liable to morphological
modification (including derivation).
M-1 Morphology : the preposition inside a can-
didate term of Noun1 Prep Noun2 structure
is equivalent to a prefix applying on Noun2:
pourrissement apr`es r´ecolte (rot after harvest)
a4 pourrissement post-r´ecolte (post-harvesting
rot);
M-2 Derivational morphology: a derivational vari-
ation that keeps the synonymy of the base
term implies a relational adjective: acidit´e du
sang (acidity of the blood) a4 acidit´e sanguine
(blood acidity). This morphosyntactic variation
could be associated with a syntactic variation:
the sequence: alimentation destin´ee `a l’homme
et `a l’animal “food destined to man and to ani-
mal” is a variation of the base-term: alimenta-
tion animale “animal food”.
Two other types of variation could have been in-
cluded in this typology: paradigmatic and anaphori-
cal variations. The first one relies on the substitution
principle of distributional linguistics (Harris, 1968).
One or two words of the base-term could be sub-
stituted by one of their synonyms without modify-
ing the syntactic structure (Hamon and Nazarenko,
2001). The second one gathers elliptical anaphora
and acronyms.
3.2 Variations reflecting conceptual
relationships
All these variations are those which could preserve
synonymy with the base term. They can, of course,
include semantic discrepancies and can refer either
to two base terms or to a base term and a con-
ceptually linked term. Thus, two different prepo-
sitions lead to two base terms: transmission par
satellite (satellite transmission) a0a1 transmission en-
tre satellites (transmission between satellites) and
internal modification (see variation S-1a) refers to
a overcomposed term: huile essentielle de sapin (fir
essence) is a hyponym of huile essentielle (essence)
and not a variation of huile de sapin (fir oil).
We propose to identify the conceptual relation-
ships betwen base terms through syntactic or mor-
phological clues. We use standard lexical functions
to express the conceptual relationships. When there
does not exist a lexical function to label a conceptual
relationship, we introduce a new lexical function (i.e
a non standard one). Standard lexical function are
written in lower-case, non-standard in upper-case.
Syntactic The internal modification of the base
structures mainly implies two types of semantic re-
lationships:
a0 Hyperonymy: if it is a relational adjective
that modifies the base term of N1 Adj or
N1 Prep (Det) N2 structure, an hyperonymic
relationship occurs between the base term
and the modified one. The lexical function
that captures hyperonymic relationships is the
function Spec introduced by (Grimes, 1990):
Spec (contraction isom´etrique “isomet-
ric contraction”) = contraction musculaire
isom´etrique “isometric muscular contraction”
Spec (agent bact´erien “bacterial agent”) =
agent infectieux bact´erien “bacterial infectious
agent”
a0 Antonymy: if it is an adverb of negation that
modifies the base term of N1 Adj structure,
an antonymic relationship occurs between the
base term and the modified one. This relation-
ship of opposition is described with the func-
tion Anti:
Anti(levure floculante “flocculating yeast”)=
levure non floculante “non-flocculating yeast”
Morphosyntactic Semantic distinctions appear
with base terms that are morphologically related to
other base terms. Two base-terms a0 a2a1a0a3a2 and a0
a4
a2
a0
a4
a2
are considered as morphologically-related if one of
the three following constraints are satisfied:
i. a0 a2 and a0
a4
a2
are head nouns and are identical.
a0a3a2 and a0
a4
a2
are expansions and are semantically
related by the use of an affix;
ii. a0 a2 and a0
a4
a2
are head nouns and are semantically
related by the use of an affix. a0a5a2 and a0
a4
a2
are
expansions and are identical;
iii. a0 a2 and a0
a4
a2
are head nouns, a0a6a2 and a0
a4
a2
are ex-
pansions, either a0 a2 and a0
a4
a2
are identical and a0 a2
and a0
a4
a2
are semantically related by the use of
a suffix such as preserved food/food preserva-
tion;
Some affixes that have been studied for French
by (Corbin, 1987) provide clues to character-
ize the semantic link occurring between two
morphologically-related candidate terms.
a0 Antonymy: the prefixes ir, d´e, non(-) applying
either on the head or expansion element on a
base term whatever is its structure characterize
an antonymic relationship. Examples are:
Anti (solubilisation micellaire “micellar
solubilization”) = insolubilisation micellaire
“micellar insolubilisation”
Anti (ph´enol polym´eris´e “polymerized phe-
nol”) = ph´enol non-polym´eris´e “unpolymerized
phenol”
a0 Set of: the suffixes age, ade applying on the
head noun of base term attest of a “set of” rela-
tionship expressed with the function Mult:
Mult (plume de canard “duck feather”) =
plumage de canards “duck feather” The two
base-terms share the same pattern.
a0 Result: A “result” relationship is expressed
with the function Na7a9a8a11a10 applying on nouns. This
relation is induced either by:
– the suffixes age, ade, erie applying on the
head noun of base terms:
Na7a9a8a12a10 (plumage de canards “duck feather”)
= plume de canard “duck feather”
Na7a9a8a12a10 (filetage du saumon “salmon fillet-
ing”) = filet de saumon “salmon fillet”;
The two base-terms share the same pat-
tern.
– or by the suffixes age, ade, erie, ment,
tion, ure associated with an inversion. We
distinguish two cases:
a13 if this morphological link involves a
N Adj structure, the function Na7a9a8a12a10 ap-
plies:
Na7a9a8a12a10 (conservation des aliments “food
preservation”) = aliment conserv´e
“preserved food”;
a13 if the morphological link involves a
N `a Vinf structure, we face a non-
standard function where the term of N
`a Vinf structure expresses the state be-
fore the process. Thus, we introduce
the new function Na14a16a15a18a17a20a19 :
Na14a21a15a22a17a20a19 (conservation des aliments “food
preservation”) = aliment `a conserver
“food to preserve”;
a0 Actor: the suffixe eur applying on the head
noun of a base term builds its actant expressed
with the function S a2 : S a2 (transport routier
“road transport” = transporteur routier “road
haulier”. The two base-terms share the same
pattern.
Other semantic relationships involving two base-
terms with the same pattern are induced by prefixes.
For those, we have to introduce new functions as:
a0 “again” relationship with the prefixes re, r´e:
AGAIN(est´erification enzymatique “enzymatic
esterification”) = r´eest´erification enzymatique
“enzymatic reesterification”;
a0 “before” relationship with the prefixe pr´e(-):
BEFORE(traitement enzymatique “enzymatic
treatment”) = pr´etraitement enzymatique “en-
zymatic pretreatment”.
4 Automatic discovery and structuring
4.1 Linguistic structuring
The term extractor program takes as input a tagged
and lemmatized corpus. The programme imple-
ments shallow parsing and morphological conflat-
ing. First, it scans the corpus, counts and extracts
strings whose syntax characterizes base-terms or
one of their variants. This collecting step uses lo-
cal grammars based on regular expressions (Abney,
1997). These grammars use the morphosyntactic
information associated with the words of the cor-
pus by the tagger. The different occurrences re-
ferring to a base term or one of its variants are
grouped as a pair formed by lemmas of the can-
didate base term. Second, morphological analysis
is performed to confluate synomymic derivational
variants of base terms such as acidit´e du sang (acid-
ity of the blood) a4 acidit´e sanguine (blood acid-
ity). Stripping-recoding morphological rules adopt
the following rule schemata:
a0
a6a2a1 a3a5a4a7a6a9a8a11a10
where:
S is the relational suffix to be deleted from the end
of an adjective. The result of this deletion is the
stem R;
M is the mutative segment to be concatenated to R
in order to form a noun.
For example, the rule [ -´e +e ] says that if there is
an adjective which ends with ´e, we should strip this
ending from it and append the string e to the stem.
The algorithm below resumes the successive steps
for identifying relational adjectives:
1. Examine each candidate of Noun Adj structure;
2. Apply a transformational rule in order to gener-
ate all the possible corresponding base nouns.
3. Search the set of candidate terms for a
pair formed with Noun1 (identical between a
Noun1 (Prep (Det)) Noun2 and a Noun1 Adj
structures) and Noun2 generated from step 2.
4. If step 3 succeeds, group the two base struc-
tures under an unique candidate term.
In Step 2, morphological rules generate one or sev-
eral nouns for a given adjective. We generate a noun
for each relational suffix class. A class of suffixes in-
cludes the allomorphic variants. This overgeneration
method used in information retrieval by (Jacquemin,
2001) gives low noise because the base noun must
not only be an attested form in the corpus, but must
also appear as an extension of a head noun.
At the end of the linguistic processing, the term
extractor proposes as output:
1. a list of pilot terms ranked from the most repre-
sentative of the corpus to the least thanks to the
Loglikelihood coefficient introduced by (Dun-
ning, 1993).
2. for each pilot term, a XML structure is pro-
vided which gathers all the base structures and
the variations encountered.
An example of such data is given in figure in Table 1.
4.2 Conceptual structuring
The conceptual structuring takes as input the data
provided by the first step. First, we present the
methodology employed to exploit variables of base
terms. We then demonstrate the labelling of concep-
tual links through morphological analysis.
4.2.1 Treatment of modification variants
In the previous step, a first list of relational ad-
jectives has been established thanks to their para-
phrasic property. (Daille, 2001) demonstrated that
candidate terms of N Adj structure where Adj is re-
lational hold a more important naming potential than
for the synonym form in N1 Prep N2. The absence
of paraphrases, the non-paraphrasability, or a com-
plex paraphrasability or a large derivational distance
between the adjective and the noun do not allow ex-
haustive identification. We extend this list by ex-
ploiting the coordination variations of N Adj base
terms. Indeed, a relational adjective holds the prop-
erty to coordinate only with other relational adjec-
tives. To summarise:
1. From industrie de l’alimentation “food indus-
try” and industrie alimentaire “food industry”,
Sorted list of candidate terms
Score Pilot term Index
785 acide gras 926
722 mise au point 344
629 acide aminer 394
559 mati`ere grasse 2002
512 r´esultat obtenir 155
472 chromatographie gazeuse 1374
469 bact´erie lactique 118
a0a1a0a1a0 a0a2a0a1a0 a0a1a0a1a0
XML structure associated to chromatographie gazeuse
a3 CAND ident=1374 freq=103
a4
a3 NPN freq=9
a4
a3 BASE
a4
a3 TERM
a4 chromatographie du gaz
a3 /TERM
a4
a3 TERM
a4 chromatographie gaz
a3 /TERM
a4
a3 TERM
a4 chromatographie de gaz
a3 /TERM
a4
a3 /BASE
a4
a3 /NPN
a4
a3 NPNA freq=80
a4
a3 BASE
a4
a3 TERM
a4 chromatographie en phase gazeuze
a3 /TERM
a4
a3 /BASE
a4
a3 MODIF
a4
a3 TERM
a4 chromatographie capillaire en phase gazeuze
a3 /TERM
a4
a3 /MODIF
a4
a3 /NPNA
a4
a3 NA freq=14
a4
a3 BASE
a4
a3 TERM
a4 chromatographie gazeuze
a3 /TERM
a4
a3 /BASE
a4
a3 /NA
a4
a3 /CAND
a4
Table 1: Output of the first step
we deduce that alimentaire is a relational ad-
jective;
2. From the coordinational variant produit agri-
cole et alimentaire “farm and food product”,
we deduce that agricole is a relational adjec-
tive.
This classic learning algorithm that is normally
bound by the number of adjectives in the corpus con-
verges in five steps. It allows to extend the set of re-
lational adjectives from 143 to 239. The following
are some examples of acquired relational adjectives:
Relational Number of Coordinated relational
adjective iterations adjectives
gazeux 1 ( microbien solide liquide
organoleptique )
ferme 2 ( ´elastique )
productif 1 ( )
global 2 ( micro-´economique
sp´ecifique local )
peroxydasique 1 ( polyph´enoloxydasique
lipoxyg´enasique
catalasique)
hydrodynamique 3 ( thermique )
Using this extended list of relational adjectives,
we automatically check all the modification variants
of collected base-terms:
a0 if a relational adjective is present, we infer
an hyperonymy link between the variant and
the base term as for contraction isom´etrique
“isometric contraction” and contraction mus-
culaire isom´etrique “isometric muscular con-
traction”, but not for organisation ordonn´ee
des mol´ecules “ordered molecule organization”
that remains a syntactic variation of organisa-
tion mol´eculaire “molecule organization”;
a0 if an adverb of negation is present, we infer an
antonymy link between the variant and the base
term as for brunissement non enzymatique “non
enzymatic browning” and brunissement enzy-
matique “enzymatic browning”.
4.2.2 Morphological conflating
To identify the conceptual relationships denoted
by derivational links, we perform a morphological
analysis using the same method as in section 4.1:
we wrote stripping-recoding morphological rules for
each conceptual relationship, we apply the overgen-
eration method and the filtering based on the pres-
ence or not of the generated base term candidates. In
order to browse the list of candidate terms, we apply
to each candidate terms successively all the possible
derivations.
The output of the conceptual structuring program
is a list of candidate terms ranked, each of them
representing a set of conceptually linked candidate
terms. An example of such structure is given in Ta-
ble 2.
5 Results and Evaluation
We apply our program on a technical corpus in the
field of agriculture which consists of 2,702 scientific
abstracts for a total of 427,482 tokens and an average
size of a record of 316 tokens.
Table 3 gives the results of the collecting phase and
Table 4 shows the percentages of the different types
of variations for candidate terms appearing at least
two times and the number of synonymic conflations.
This conflating has a linguistic precision of 99 %.
Number of occurrences 1 a0 2 Total
base structures
Nom1 Prep (Det) Nom2 17 232 5 949 23 181
Nom Adj 12 344 4 778 17 122
Nom `a Vinf 203 16 219
Total 29 912 10 895 40 807
Table 3: Number of candidate base terms
Syntactic variation
Coor + Modif Coor Modif
61 (0,5 %) 458 (4 %) 1651 (15,1 %)
19,2 %
Morphological variation
N1 (Prep (Det)) N2 / N1 AdjR
with AdjR derived from N2
343
Table 4: Number of base-term variations
Conceptual Link Syntactic Morphological
Spec 731
Anti 183 106
Na1a3a2a5a4 132
MICRO 59
AGAIN 36
BEFORE 29
Mult 23
INTER 20
a0a1a0 a0 a0 a0a2a0
Total 914 558
Table 5: Number of major conceptual links
Table 5 gives the number of the most frequent auto-
matically identified conceptual relationships.
Concerning morphological links, we note that two
non-standard functions that have not been presented
yet obtain a consequent representativity: MICRO
induces from the suffixe micro(-): MICRO(film per-
for´e “perforated film”) = film micro-perfor´e “pin-
hole film”; INTER infers from the suffixe inter ex-
pressing a reprocivity relationship: INTER(´echelle
nationale “national scale”) = ´echelle internationale
“international scale”. The average precision of
the morphological links is 95 %. The wrong
links are 75 % due to the prefixes re, r´e refering
to the function AGAIN. Examples of false drop
are: action/r´eaction “reaction” in several candidate
base-terms: action/r´eaction enzimatique “enzimatic
action/reaction”, action/r´eaction acide “acide ac-
tion/reaction”, a0a1a0a1a0 , production/reproduction, solu-
tion/r´esolution “resolution”, etc.
6 Conclusion
Links between complex terms can be used to as-
sist terminolographers to handle extracted termino-
logical data more conveniently, since several related
concepts are clustered. This conceptual structuring
relies on term variation and we have stressed the cru-
cial part of this handling. The method can be eas-
ily adapted to other romance languages and English
for which it suffices to define patterns for base-terms
and their variations and to list appropriate affixes re-
flecting semantic relationships.

References
Steven Abney. 1997. Part-of-Speech Tagging and Partial
Parsing. In Steve Young and Gerrit Bloothooft, edi-
tors, Corpus-Based Methods in Language and Speech
Processing, volume 2, chapter 4. Kluwer Academic
Publishers.
M. Teresa Cabr´e, Rosa Estop`a Bagot, and Jordi Vivaldi
Platresi. 2001. Automatic term detection: A re-
view of current systems. In Didier Bourigault, Chris-
tian Jacquemin, and Marie-Claude L’Homme, editors,
Recent Advances in Computational Terminology, vol-
ume 2 of Natural Language Processing, pages 53–88.
John Benjamins.
Danielle Corbin. 1987. Morphologie d´erivationnelle
et structuration du lexique. Max Niemeyer Verlag
T¨ubingen.
Judit Feliu Cort`es and M. Teresa Cabr´e. 2002. Concep-
tual relations in specializes texts: new typology and an
extraction system proposal. In Proceeding of the 6th
International Conference of Terminology and Knowl-
egde Engineering (TKE’02).
B´eatrice Daille. 2001. Qualitative terminology extrac-
tion. In Didier Bourigault, Christian Jacquemin, and
Marie-Claude L’Homme, editors, Recent Advances
in Computational Terminology, volume 2 of Natu-
ral Language Processing, pages 149–166. John Ben-
jamins.
Ted Dunning. 1993. Accurate methods for the statistics
of surprise and coincidence. Computational Linguis-
tics, 19(1):61–74.
Natalia Grabar and Pierre Zweigenbaum. 2000. Au-
tomatic acquisition of domain-specific morphological
resources from thesauri. In In Proceedings of RIAO
2000: Content-Based Multimedia Information Access,
pages 765–784, Paris, France.
J. Grimes. 1990. Inverse lexical functions. In J. Steele,
editor, Meaning-Text Theory: Linguistics, Lexicogra-
phy and Implications, pages 350–364. Ottawa Univer-
sity Press, Ottawa.
Thierry Hamon and Adeline Nazarenko. 2001. Detection
of synonymy link between terms: Experiment and re-
sults. In Didier Bourigault, Christian Jacquemin, and
Marie-Claude L’Homme, editors, Recent Advances
in Computational Terminology, volume 2 of Natu-
ral Language Processing, pages 185–208. John Ben-
jamins.
Zelig S. Harris. 1968. Mathematical Structures of Lan-
guage. Wiley, New York.
C. Jacquemin. 2001. Spotting and Discovering Terms
through Natural Language Processing. Cambridge:
MIT Press.
Kyo Kageura. 2002. The Dynamics of Terminology:
A Theoretico-Descriptive Study of Term Formation
and Terminological Growth, volume 5 of Terminology
and Lexicography Research and Practice. John Ben-
jamins.
M.-C. L’Homme. 2002. Fonctions lexicales pour
repr´esenter les relations smantiques entre termes.
Traitement automatique des langues (TAL), 43(1):19–
42.
Emmanuel Morin and Christian Jacquemin. 1999. Pro-
jecting corpus-based semantic links on a thesaurus. In
Proceedings of the 37th Annual Meeting of the Associ-
ation for Computational Linguistics (ACL’99), pages
389–396.
J.C. Sager. 1990. A Practical Course in Terminology
Processing. John Benjamins.
Leo Wanner, editor. 1996. Lexical Functions in Lexicog-
raphy and Natural Language Processing. John Ben-
jamins, Amsterdam/Philadelphia.
