Modeling Monolingual and Bilingual Collocation Dictionaries in
Description Logics
Dennis Spohr and Ulrich Heid
Institute for Natural Language Processing
University of Stuttgart
Azenbergstr. 12, D-70174 Stuttgart, Germany
{spohrds,heid}@ims.uni-stuttgart.de
Abstract
This paper discusses an approach to mod-
eling monolingual and bilingual dictio-
naries in the description logic species of
the OWL Web Ontology Language (OWL
DL). The central idea is that the model
of a bilingual dictionary is a combination
of the models of two monolingual dictio-
naries, in addition to an abstract transla-
tion model. The paper addresses the ad-
vantages of using OWL DL for the design
of monolingual and bilingual dictionaries
and proposes a generalized architecture for
that purpose. Moreover, mechanisms for
querying and checking the consistency of
such models are presented, and it is con-
cluded that DL provides means which are
capable of adequately meeting the require-
ments on the design of multilingual dictio-
naries.
1 Introduction
We discuss the modeling of linguistic knowledge
about collocations, for monolingual and bilingual
electronic dictionaries, for multiple uses in NLP
and for humans.
Our notion of collocation is a lexicographic
one, inspired by (Bartsch, 2004); we start from
her working definition: “collocations are lexically
and/or pragmatically constrained recurrent cooc-
currences of at least two lexical items which are
in a direct relation with each other.” The fact of
being lexically and/or pragmatically constrained
leads to translation problems, as such constraints
are language specific. With Hausmann (2004), we
assume that collocations have a base and a col-
locate, where the base is autosemantic and thus
translatable without reference to the collocation,
whereas the collocate is synsemantic, i.e. its read-
ing is selected within a given collocation. Ex-
amples of collocations according to this defini-
tion include adjective+noun-combinations (heavy
smoker, strong tea, etc.), verb+subject- (question
arises, question comes up) and verb+complement-
groups (give+talk, take+walk) etc. The definition
excludes however named entities (Rio de Janeiro)
and frequent compositional groups (e.g. the police
said...). Our data have been semi-automatically
extracted from 200 million words of German
newspaper text of the 1990s (cf. Ritz (2005)).
We claim that a detailed monolingual descrip-
tion of the linguistic properties of collocations pro-
vides a solid basis for bilingual collocation dictio-
naries. The types of linguistic information needed
for NLP and those required for human use, e.g. in
text production or translation into a foreign lan-
guage, overlap to a large extent. Thus it is rea-
sonable to define comprehensive monolingual data
models and to relate these with a view to transla-
tion.
In section 2, we briefly list the most impor-
tant phenomena to be captured (see also Heid and
Gouws (2006)); section 3 introduces OWL DL,
motivates its choice as a representation format and
describes our monolingual modeling. In section 4,
we discuss and illustrate the bilingual dictionary
architecture.
2 Collocation Data
Properties of collocations. A mere list of word
pairs or sequences (give a talk, lose one’s patience)
is not a collocation dictionary. For use in NLP, lin-
guistic properties of the collocations and of their
components must be provided: these include the
category of the components (giveV + talkN), the
65
distribution of base (talk) and collocate (give), as
well as morphosyntactic preferences, e.g. with re-
spect to the number of an element (e.g. have high
hopes), the use of a determiner (lose one’sposs|{}
patience, cf. Evert et al. (2004)).
For collocations to be identifiable in the context
of a sentence (e.g. to avoid attachment ambiguity
in parsing) and, conversely, in generation, to be
correctly inserted into a sentence, the syntagmatic
behavior of collocations must be described. This
includes their function within a sentence (e.g in
the case of adverbial NPs) and the subcategoriza-
tion of their components, e.g. with support verb
constructions (make the proposal to + INF). As
subcategorization is not fully predictable from
the subcategorization of the noun (how to explain
the preposition choice in Unterst¨utzung finden
bei jmdm, ‘find support in so.’, be supported?),
we prefer to encode the respective data in the
monolingual dictionary. To support translation
mapping at the complement level, the representa-
tion of each complement contains its grammatical
category (NP, AP, etc.), its grammatical func-
tion (subject, object, etc.) and a semantic role
inspired by FrameNet1. This allows us to cater
for divergence cases: jmdSubj/SPEAKER bringt
jmdmInd.Obj/ADDRESSEE etw.Obj/TOPIC in
Erinnerung vs. someoneSubj/SPEAKER reminds
someoneObj/ADDRESSEE of sth.Prep.Obj/TOPIC.
Relations involving collocations. For language
generation, paraphrasing or for summarization,
paradigmatic relations of collocations must also
be modeled. These include synonymy, antonymy
and taxonomic relations, but also morphological
ones (word formation) and combinations of col-
locations. Synonymy and antonymy should re-
late collocations with other collocations, but also
with single words and with idioms: all three types
should have the same status. Next to strict syn-
onymy, there may be ‘quasi-synonymy’.
Transparent noun compounds tend to share col-
locates with their heads (Pause einlegen, Rauch-
pause einlegen, Kaffeepause einlegen): if the re-
lation between compound and head (Kaffeepause
– Pause) and between the respective collocations
is made explicit, this knowledge can be exploited
in translation, when a compositional equivalent
is chosen (have a (smoking/coffee) break). Para-
phrasing and its applications also profit from an
explicit representation of morphological relations
1Cf. http://framenet.icsi.berkeley.edu/
between collocates: submit + proposal, submis-
sion of + proposal and submitter of + proposal all
refer to the same collocational pattern.
A formal model for a collocation dictionary,
monolingual and/or bilingual, has to keep track
of the above mentioned properties and relations of
collocations; both should be queriable, alone and
in arbitrary combinations.
Other collocation dictionaries and dictionary
architectures. Most of the above mentioned
properties and relations have been discussed in the
descriptive literature, but to our knowledge, they
have never been modeled all in an electronic dic-
tionary. The Danish STO dictionary (Braasch and
Olsen, 2000) and Krenn’s (2000) database of Ger-
man support verb+PP-constructions both empha-
size morphosyntactic preferences, but do not in-
clude relations. The electronic learners’ dictionar-
ies DAFLES and DICE2 focus on semantic expla-
nations of collocations, but do not contain details
about most of the properties and relations men-
tioned above. The implementation of Mel’ˇcuk’s
Meaning⇔Text-Theory in the DiCo/LAF model3
comes closest to our requirements, insofar as it
is highly relational and includes some though not
all of the morphological relations we described
above.
The Papillon project (S´erasset and Mangeot-
Lerebours, 2001) proposes a general architecture
for the interlingual linking of monolingual dictio-
naries; as it is inspired by the DiCo formalizar-
ion, it foresees links between readings, e.g. to ac-
count for morphological relations. This mecha-
nism could in principle be extended to syntag-
matic phenomena; we are, however, not aware of
a Papillon-based collocation dictionary.
3 Modeling in OWL DL
In this section, we present the main features of
OWL DL and their relevance to the modeling of
lexical data. Section 3.2 addresses the design of
a monolingual collocation dictionary using OWL
DL (Spohr, 2005).
3.1 Main Features of OWL
OWL DL is the description logic sublanguage
of the OWL Web Ontology Language (Bech-
2Cf. http://www.kuleuven.ac.be/dafles/
and DICE: http://www.dicesp.com/
3Cf. http://olst.ling.umontreal.ca/
dicouebe/
66
hofer et al., 2004), combining the expressivity of
OWL with the computational completeness and
decidability of Description Logics (Baader et al.,
2003)4. Properties of OWL DL relevant for lexical
modeling are listed and discussed in the following.
Classes. An OWL DL data model consists of
a subsumption hierarchy of classes, i.e. a class
X subsumes all its subclasses X1 to Xn. While
classes represent concepts, their instances (called
OWL individuals) represent concrete manifesta-
tions in the model. Classes and their instances can
be constrained by stating assertions in the model
definition, e.g. a class can be defined as being
disjoint with other classes, which means that in-
stances of a certain class cannot at the same time
be instances of the disjoints of this particular class.
Properties. Classes are described by properties.
These can be used either to specify XML Schema
Datatypes (datatype properties) or to relate in-
stances of one class to instances of (probably)
other classes (object properties). These classes are
then defined as the domain and range of a property,
i.e. a particular property may only relate instances
of classes in its domain to instances of classes in
its range. In addition to this, a property may be
assigned several distinct formal attributes, such as
symmetric, transitive or functional, and can be de-
fined as the inverse of another property. Similar
to classes, properties can be structured hierarchi-
cally as well, which, among others, facilitates the
use of underspecified information in queries (see
section 3.2).
Inferences. The possibility to infer explicit
knowledge from implicit statements is a core fea-
ture of OWL DL and can be performed by using
DL reasoners (such as FaCT5, Pellet6 or Racer-
Pro7). The most basic inference is achieved via
the subsumption relation among classes or prop-
erties in the respective hierarchy (see above), but
also more sophisticated inferences are possible.
Among others, these may involve the formal at-
tributes of properties just mentioned. For example,
4As the emphasis in our work is on morphology, syntax
and lexical combinatorics, we profit from the formal prop-
erties of DL without feeling the need for non-monotonicity
as implemented, for example, in DATR (Evans and Gazdar,
1996).
5http://www.cs.man.ac.uk/˜horrocks/
FaCT/
6http://www.mindswap.org/2003/pellet/
7http://www.racer-systems.com
stating that instance A is linked to B via a sym-
metric property P leads a reasoner to infer that B
is also linked to A via P. In conjunction with tran-
sitivity, a relatively small set of explicit statements
may suffice to interrelate several instances implic-
itly (i.e. all instances in a particular equivalence
class created by P).
Consistency. In addition to inferences, DL rea-
soners can further be used to check the consistency
of an OWL DL model. One of the primary ob-
jectives is to check whether the assertions made
about classes and their instances (see above) are
logically consistent or whether there are contradic-
tions. This consistency checking is based on the
open-world assumption, which states that “what
cannot be proven to be true is not believed to be
false” (Haarslev and M¨oller, 2005). Since lexi-
cal data occasionally demand a closed world, other
checking formalisms are required, which are men-
tioned in section 3.2 below.
3.2 Monolingual Collocation Dictionary
A data model for a monolingual collocation dic-
tionary based on OWL DL has been presented
in (Spohr, 2005). It was designed using the
Prot´eg´e OWL Plugin (Knublauch et al., 2004) and
makes use of the advantages of OWL DL men-
tioned above.
Lexical vs. descriptive entities. On the class
level, the model distinguishes between lexical en-
tities (e.g. single-word and multi-word entities,
such as collocations or idioms) and descriptive en-
tities (e.g. gender, part-of-speech, or subcategori-
sation frames), with lexical entities being linked
to descriptive entities via properties. More than 40
of these descriptive properties have been modeled.
In order to reflect the distinction between metalan-
guage vocabulary and object language vocabulary,
the two types of entities can be separated such that
they are part of different models. In other words,
the classes and instances of descriptive entities
constitute a model of descriptions, which is im-
ported by a lexicon model containing classes and
instances of lexical entities (see also section 4.1
below).
Lexical relations. In addition to descrip-
tive properties, the data model also contains
a number of lexical relations linking lexical
entities, such as morphological or semantic
relations. These relations have been structured
67
hierarchically and contain several subproperties,
such as hasCompound or isSynonymOf,
which use the formal attributes mentioned in
section 3.1. For instance, isSynonymOf
has been defined as a symmetric and transi-
tive property (as opposed to the non-transitive
isQuasiSynonymOf; see section 2), while
hasCompound has been defined as the inverse of
a property isCompoundOf. A small sample of
descriptive and lexical relations of the collocation
Kritik ¨uben is illustrated in Figure 1 below.
Property Value
hasLemma “Kritik ¨uben”
hasCompound Selbstkritik ueben
isSynonymOf kritisieren VV 1
hasCollocationType V-Nobj acc
hasComplementation SubcatFrame 12
hasExampleSentence Example 84
isInCorpus HGC-STZ
Figure 1: Sample of the properties of Kritik ¨uben
Semantic relations link lexical entities on
the conceptual (i.e. word sense) level. There-
fore, the synonym of Kritik ¨uben is not some
general single-word entity kritisieren VV,
but a particular word sense of kritisieren,
kritisieren VV 1 in this case (see
Spohr (2005) for more detail).
Queries. The data model can be queried very
efficiently using the Sesame framework (Broek-
stra et al., 2002; Broekstra, 2005) and its associ-
ated query language SeRQL. An example query
retrieving all collocations and their types is given
below, along with a sample of the results8.
SELECT *
FROM {A} rdf:type {lex:Collocation},
{A} lex:hasCollocationType {B}
A B
in_Frage_kommen V-PPpobj
Kritik_ueben V-Nobj_acc
Lob_aussprechen V-Nobj_acc
zu_Last_legen V-PPpobj
Figure 2: Query for retrieving collocations and
their types, along with results
Due to the fact that the relations in the data
8In these examples, lex: is the namespace prefix for re-
sources defined in the data model.
model have been structured hierarchically, it is
possible to state underspecified queries. Figure 3
illustrates an underspecified query for semanti-
cally related entities, regardless of the precise na-
ture of this relation. Hence, the first two rows
in the result table below contain synonym pairs,
while the last two rows contain antonym pairs.
SELECT *
FROM {A} lex:hasSemanticRelationTo {B}
A B
Kritik_ueben kritisieren_VV_1
kritisieren_VV_1 Kritik_ueben
Kritik_ueben Lob_aussprechen
Lob_aussprechen Kritik_ueben
Figure 3: Underspecified query for semantically
related entities, along with results
As is indicated in Figure 3, the results appear
twice, i.e. they contain every combination of those
entities between which the relation holds. This is
due to the fact that the respective semantic rela-
tions have been defined as symmetric properties
(see above).
Consistency and data integrity. Section 3.1
mentioned the distinction between the open-world
assumption and the closed-world assumption.
While the consistency checking performed by DL
reasoners is generally based on an open world, it
is vital especially for lexical data to simulate a
closed world in order to check data integrity. Con-
sider, for instance, the assertion that every collo-
cation has to have a base and a collocate. Due to
the open-world assumption, a DL reasoner would
never render a collocation violating this constraint
inconsistent, simply because it cannot prove that
this collocation has either no base or no collo-
cate. In order for this to happen, the simulation
of a closed world is needed. In our approach, this
is achieved by stating consistency constraints in
SeRQL. Figure 4 below illustrates a constraint for
the purpose just mentioned.
This query retrieves all collocations and sub-
tracts those who have a path to both a base and
a collocate. The result set then contains exactly
those instances which have either no base or no
collocate.
68
SELECT Coll
FROM {Coll} rdf:type {lex:Collocation}
MINUS
SELECT Coll
FROM {Coll} lex:hasBase {};
lex:hasCollocate {}
Figure 4: Constraint checking: does every collo-
cation have a base and a collocate?
4 Bilingual Model Architecture
Based on the definition of a monolingual colloca-
tion dictionary described above, the architecture of
a bilingual dictionary model can be designed such
that it is made up of several components (i.e. OWL
models). These are introduced in the following.
4.1 Components of a Bilingual Dictionary
The components of a bilingual dictionary are illus-
trated in Figure 5.
Translation model
Bilingual dictionary model
createdFrom
Monolingual dictionary model
Model of descriptions
Lexicon model
imports
createdFromcreatedFrom
imports imports
Monolingual dictionary model
Figure 5: Architecture of a bilingual dictionary
model
Model of descriptions. The most basic com-
ponent of a bilingual dictionary model is a
model of descriptions, which contains language-
independent classes and instances of descriptive
entities, as well as the relations among them (see
section 3.2).
Lexicon model. The model of descriptions is
imported by an abstract lexicon model via the
owl:imports statement (see (Bechhofer et al.,
2004)). The effect of using the import statement
is that the lexicon model can access the classes,
instances and properties defined in the description
model without being able to alter the data therein.
In addition to the thus available classes, the lexi-
con model further provides classes of lexical enti-
ties and relations among them, as well as relations
linking lexical and descriptive entities.
Monolingual dictionary model. The lexicon
model serves as input for the creation of a mono-
lingual dictionary model, i.e. the lexicon model is
not imported by the dictionary model, rather the
dictionary model is an instantiation of it. There are
practical reasons for doing so, the most important
one being that the class of lexical entities (defined
in the lexicon model) and its instances (defined
in the monolingual dictionary) thus have the same
namespace prefix, which would not be the case if
the lexicon model was imported by the monolin-
gual dictionary. The advantages are most obvious
in the context of the mapping between monolin-
gual dictionary models (see section 4.2). Finally,
a monolingual dictionary may further introduce its
own instances (or even classes) of descriptive en-
tities, i.e. descriptions which are language-specific
and which are hence not part of the language-
independent model of descriptions (see above).
Translation model. The translation model is an
abstract model containing only relations between
monolingual dictionary models, i.e. it does not
contain class definitions. Since the model is re-
quired to be generic, these relations do not have
a specified domain and range, as otherwise the
translation model would be restricted to a single
language pair. The specification of the domain
and range of the relations is performed in the fi-
nal model of the bilingual dictionary.
Bilingual dictionary model. The bilingual dic-
tionary model is an instantiation of the translation
model. It further imports two monolingual dictio-
69
nary models and specifies the domain and range of
the abstract relations in the translation model (see
section 4.2 below).
4.2 Mapping between Models
By importing the monolingual dictionaries, each
of these models is assigned a unique namespace
prefix, e.g. english: or german:. Thus, in
an English-German dictionary, for instance, a rela-
tion calledhasTranslationmay be defined as
a symmetric property linking lexical entities of the
English monolingual dictionary model (i.e. its do-
main is defined for instances with the english:
prefix) to lexical entities of the German model
(i.e. instances with german:). This translation
mapping is illustrated in Figure 6 for the colloca-
tion Kritik ¨uben.
express criticism
MWE: Collocation Single−Word Entity
criticize
Monolingual English Dictionary Model
Paraphrase
"to criticize very fiercely"
Kritik üben kritisieren
MWE: Collocation Single−Word Entity
MWE: Idiom
in der Luft zerreißen
Monolingual German Dictionary Model
Figure 6: Translation mapping between monolin-
gual dictionaries
As is indicated there, multi-word entities can be
translated as single-word entities and vice versa.
Moreover, since hasTranslation has been
defined as a symmetric property, the translation
mapping is bidirectional. However, since some in-
stance in one language model might not have an
equivalent instance in the other model, a further
property can be defined which links the respective
entity to a new instance created in the bilingual
model (see Paraphrase in the figure above). As
this instance is only required for the modeling of
this particular bilingual dictionary, it is not part of
the “original” monolingual models, and hence the
relation between the respective entities is not bidi-
rectional.
In addition to the translation mapping of lexi-
cal entities, it may further be necessary to map
instances of descriptive entities of one model
onto instances in the other model. As was men-
tioned in section 4.1, the model of descriptions
contains language-independent descriptive enti-
ties. Since both monolingual dictionaries import
the model of descriptions (via the lexicon model),
the two “versions” of it are unified in the bilin-
gual model. However, it is certainly conceivable
to have two languages which both avail them-
selves of a descriptive entity that is not language-
independent, but which is the same for the two
languages in question. For example, not all lan-
guages have the gender neuter. English and
German, however, do have it, and therefore an
English-German bilingual dictionary has to ex-
press that english:neuter is the same as
german:neuter. In OWL, this can be achieved
by using the owl:sameAs statement, which ex-
presses exactly the circumstances just mentioned.
4.3 Example Query
A query retrieving the situation depicted in
Figure 6 is given below. It extracts the
(quasi-)synonyms of Kritik ¨uben (which Kritik
¨uben itself is a part of) and their respective transla-
tions and/or paraphrases. The latter is achieved by
restricting the properties that Rel2 may stand for
to those having the prefix bdm:, i.e. the prefix de-
fined for the bilingual dictionary model. In other
words, the query leaves the exact relation between
B and C underspecified and simply restricts it to
being defined in the bilingual dictionary, which
only contains relations linking instances belonging
to different monolingual dictionaries. The results
are shown in the table below.
5 Conclusion
We have described a model for monolingual and
bilingual collocation dictionaries in OWL DL.
This formalism is well suited for the intended
modularization of linguistic resources, be they
language- or language-pair- specific (our dictio-
70
SELECT DISTINCT B, C
FROM {} german:hasLemma {A};
Rel1 {B} Rel2 {C}
WHERE A LIKE "Kritik ¨uben"
AND (Rel1 = german:isSynonymOf
OR Rel1 = german:isQuasiSynonymOf)
AND namespace(Rel2) = bdm:
B C
kritisieren_VV_1 express_criticism
kritisieren_VV_1 criticize_VV_1
Kritik_ueben express_criticism
Kritik_ueben criticize_VV_1
in_Luft_zerreissen “to criticize very fiercely”
Figure 7: Query for retrieving the
(quasi-)synonyms of Kritik ¨uben and their
translations and paraphrases, along with results
nary models), generalized over one or more lan-
guages (our lexicon model), or more abstract, in
the sense of a meta-model or an inventory of the
descriptive devices shared by the linguistic de-
scriptions of several languages (our model of de-
scriptions, see figure 5 above). This model of
descriptions will be larger for related languages
(e.g. the indo-european ones), and smaller for ty-
pologically very diverse languages; it is however
by no means meant to have any interlingual, let
alone universal function, but is rather understood
in the sense of PARGRAM’s shared inventory of
descriptive devices9.
We have modelled so far about 1000 colloca-
tions, their components, preferences and relations
(also with single words); we intend to consider-
ably enlarge the collocation dictionary, using the
possibilities to combine OWL DL models with
databases, offered by the Sesame framework. The
formalism also supports experiments with credu-
lous inferencing at the level of translation equiv-
alence, e.g. by following not only explicit equiv-
alence relations, but also synonymy relations: in
line with the query discussed in section 4.3 above
(cf. Figure 7), one could also start from the English
express criticism and retrieve the equivalent collo-
cation Kritik ¨uben as well as its (quasi-)synonyms
kritisieren (single word) and in der Luft zerreißen
(idiom), which may thus be proposed as equivalent
candidates for express criticism.
More such investigations into the data collec-
9Cf. http://www2.parc.com/istl/groups/
nltt/pargram/gram.html
tion are planned; they may require non-standard
access to the dictionary, i.e. access via paths in-
volving other properties and relations than just
lemmas and equivalents. The relational nature of
the dictionary supports this kind of exploration;
we intend to specify and implement a ‘linguist-
friendly’ query overlay to SeRQL and a Graphi-
cal User Interface to make such explorations more
easy.

References
Franz Baader, Diego Calvanese, Deborah L. McGuin-
ness, Daniele Nardi, and Peter F. Patel-Schneider.
2003. The Description Logic Handbook: Theory,
Implementation and Applications. Cambridge Uni-
versity Press, Cambridge, UK.
Sabine Bartsch. 2004. Structural and Functional
Properties of Collocations in English. A Corpus
Study of Lexical and Pragmatic Constraints on Lex-
ical Cooccurrence. Narr, T¨ubingen, Germany.
Sean Bechhofer, Frank van Harmelen, Jim Hendler, Ian
Horrocks, Deborah L. McGuinness, Peter F. Patel-
Schneider, and Lynn Andrea Stein. 2004. OWL
Web Ontology Language Reference. Technical re-
port.
Anna Braasch and Sussi Olsen. 2000. Formalised
representation of collocations in a Danish compu-
tational lexicon. In Proceedings of the EURALEX
International Congress 2000, Stuttgart, Germany.
Jeen Broekstra, Arjohn Kampman, and Frank van Her-
melen. 2002. Sesame: A Generic Architecture for
Storing and Querying RDF and RDF Schema. In
Proceedings of the First International Semantic Web
Conference (ISWC 2002), pages 54–68, Sardinia,
Italy.
Jeen Broekstra. 2005. Storage, Querying and Infer-
encing for Semantic Web Languages. Ph.D. thesis,
Vrije Universiteit Amsterdam, The Netherlands.
Roger Evans and Gerald Gazdar. 1996. DATR: A lan-
guage for lexical knowledge representation. Com-
putational Linguistics, 22(2):167–216.
Stefan Evert, Ulrich Heid, and Kristina Spranger.
2004. Identifying Morphosyntactic Preferences in
Collocations. In Proceedings of LREC-2004, Lis-
bon, Portugal.
Volker Haarslev and Ralf M¨oller, 2005. RacerPro
User’s Guide and Reference Manual, Version 1.8.1.
Franz Josef Hausmann. 2004. Was sind eigentlich Kol-
lokationen? In Karin Steyer, editor, Wortverbindun-
gen - mehr oder weniger fest, pages 309–334. Insti-
tut f¨ur Deutsche Sprache: Jahrbuch 2003.
Ulrich Heid and Rufus H. Gouws. 2006. A model
for a multifunctional electronic dictionary of collo-
cations. Draft of a paper submitted to EURALEX
2006.
Holger Knublauch, Mark A. Musen, and Alan L. Rec-
tor. 2004. Editing description logic ontologies
with the Prot´eg´e OWL plugin. In Proceedings of
the International Workshop in Description Logics -
DL2004, Whistler, BC, Canada.
Brigitte Krenn. 2000. The Usual Suspects: Data-
Oriented Models for Identification and Representa-
tion of Lexical Collocations. Ph.D. thesis, DFKI
Universit¨at des Saarlandes, Saarbr¨ucken, Germany.
Julia Ritz. 2005. Entwicklung eines Systems
zur Extraktion von Kollokationen mittels mor-
phosyntaktischer Features. Diploma thesis, Insti-
tut f¨ur Maschinelle Sprachverarbeitung, Universit¨at
Stuttgart, Germany.
Gilles S´erasset and Mathieu Mangeot-Lerebours.
2001. Papillon lexical database project: Monolin-
gual dictionaries & interlingual links. In Proceed-
ings of the Sixth Natural Language Processing Pa-
cific Rim Symposium: NLPRS-2001, pages 119–125,
Tokyo, Japan.
Dennis Spohr. 2005. A Description Logic Approach
to Modelling Collocations. Diploma thesis, Insti-
tut f¨ur Maschinelle Sprachverarbeitung, Universit¨at
Stuttgart, Germany.
