=Paper=
{{Paper
|id=Vol-465/paper-2
|storemode=property
|title=Legal Taxonomy Syllabus version 2.0
|pdfUrl=https://ceur-ws.org/Vol-465/paper2.pdf
|volume=Vol-465
}}
==Legal Taxonomy Syllabus version 2.0==
https://ceur-ws.org/Vol-465/paper2.pdf
Legal Taxonomy Syllabus version 2.0
Gianmaria Ajani1 , Guido Boella2 , Leonardo Lesmo2 , Marco Martin2 ,
Alessandro Mazzei2 , Daniele P. Radicioni2 , and Piercarlo Rossi3
1
Dipartimento di Scienze Giuridiche - Università di Torino
2
Dipartimento di Informatica - Università di Torino
3
Dipartimento di Studi per l’Impresa e il Territorio - Università del Piemonte
Orientale
gianmaria.ajani@unito.it, {guido,lesmo,mazzei,radicion}@di.unito.it,
notmart@gmail.com, piercarlo.rossi@eco.unipmn.it
Abstract. The need for managing the conceptual representation of Eu-
ropean law led to the development of the Legal Taxonomy Syllabus (LTS)
and the related methodology. In this paper we consider further legal is-
sues that emerged during the test and use phases, and outline the new
features that we added to the new version, the LTS 2.0.
1 Introduction
European Union Directives (EUDs) are sets of norms that have to be imple-
mented by the national legislations and translated into the language of each
Member State. The general problem of multilingualism in European legislation
has recently been tackled by linguistic and ontological tools [8,5,17,18]. The man-
agement of EUD is particularly complex, since the implementation of a EUD does
not correspond to a straight transposition into a national law.
In previous work we carried out the Legal Taxonomy Syllabus4 (LTS), a tool
to build multilingual conceptual dictionaries aimed at representing and analysing
terminologies and concepts from EUDs [1,2]. LTS is based on the distinction be-
tween terms and concepts. The latter ones are arranged into ontologies that are
organised in levels. Only two levels were defined: the European level –containing
only one ontology deriving from EUDs annotations–, and the national level
–hosting the distinct ontologies deriving from the legislations of EU member
states.
While annotating the EUDs, testing and using the system, some more re-
quirements emerged from users expert in law, demanding for a more sophis-
ticated approach along with further developmental efforts: first, it is frequent
the case of concepts which are the result of a doctrinal interpretation process
rather than of the definition in directives. If, on the one hand, the definitions in
directives and their relation with the actual text are required by legal scholars
to have a precise model of European law, the layman is more interested in the
4
http://www.eulawtaxonomy.org
concepts which results from the doctrinal interpretation. Furthermore, laws are
typical objects evolving through time. An open issue to cope with in building
legal frameworks both at the European and at the national level is the norma-
tive change [12,4]. Concepts in the legal ontologies should not only represent the
consolidated legal text, but should also keep trace of the evolution of meaning.
In this paper we consider not only the terms defined in the directives, but
also the interpretation process of legal scholars in the LTSand how to better in-
tegrate concepts and the text of EUDs in the LTS. We answer the first question
by introducing abstract concepts (abstract in that they are not related to a single
directive), which should be conveniently recognized as a grouping of concepts.
The users will be thereby allowed to navigate the ontology at different levels
of details depending on their goals. Moreover, exploiting natural language pro-
cessing techniques we greatly simplify the management of legal text associated
to concepts. Also, we investigate how to extend the ontology with a temporal
dimension to the ends of representing normative change, and to allow users to
search also for past meanings of terms and the modified norms introducing them.
To these ends, we introduce time into the ontology, and allow new concepts to
replace the old ones while keeping the latter ones in the system as well.
2 Multilingual and Multilevel Ontologies for European
Directives
Comparative Law has identified two key points in dealing with EUD, which
make more difficult dealing with the polysemy of legal terms. We call them the
terminological and conceptual misalignments. The first problem is determined by
the lexical ambiguity of the legal terms (in particular homonymy) in the trans-
lation of EUDs. The second problem is determined by the lexical and conceptual
ambiguity of the legal terms (in particular polysemy) in the implementation of
EUDs. These issues determined the development of the first release of the LTS,
and have been illustrated in [2].
We now illustrate further issues in handling EUDs that required to devise
further features to enrich the original LTS.
2.1 Concepts Abstraction
The LTS system relies on the concept of unitary-meaning or umeaning: such
atomic concepts can be derived from excerpts of the text of legal norms, such
as European directives or national laws, and are arranged into two separate
categories of umeanings, as described in [2]. EUDs provide rigorous definitions
of some terms, such as the definition of the Italian term consumatore (consumer ),
in the Italian version of the EUD 93/13/EEC, Art. 2 is:
[. . . ](b) “consumatore”: qualsiasi persona fisica che, nei contratti oggetto della
presente direttiva, agisce per fini che non rientrano nel quadro della sua attività
professionale; [...]
[. . . ](b) “consumer”: means any natural person who, in contracts covered by
this Directive, is acting for purposes which are outside his professional activity;
[. . . ] (our literal translation)
However, two facts must be pointed out. Different EUDs might affect different
aspects of the legislation: thus the definition of a term in a EUD only applies
to a specific context. Furthermore, EUDs could be written at different points in
time, and they can introduce diverging definitions. Let us consider the definition
of consumatore, as it appears in the Italian version of the EUD 2002/65/EC,
Art. 1:
[. . . ](d) “consumatore”: qualunque persona fisica che, nei contratti a distanza,
agisca per fini che non rientrano nel quadro della propria attività commerciale
o professionale; [. . . ]
[. . . ](d) “consumer”: means any natural person who, in distance contracts
covered by this Directive, is acting for purposes which are outside his business
or professional activity; [. . . ] (our literal translation)
We remark that in contrast with English, in Italian the second definition of con-
sumatore is broader than the first one, since the term professionale (professional )
does not include commerciale (business). This divergence of term definitions can
often occur, since EUDs have usually a sectorial specific target. In this way, EUDs
covering different sectors can provide different definitions, and as many views on
the same concept. Lawyers and legislators started to put together highly sec-
torial concepts into more abstract concepts with broader meaning, in order to
describe (complex) entities, such as the consumatore in all of its aspects.
In recent years, in the Italian legislation EUDs are not being implemented
as single laws, but rather as groups of EUDs. The juridical concepts are defined
as the union of all the sectorial concepts provided by the individual EUDs, as
a result of the doctrinal interpretation process of directives. These problems
are common to all European languages. Consider, for instance the definition of
consumer, in the English version of the EUD 1999/44/EC, Art. 1.2 is:
[. . . ] (a) consumer: shall mean any natural person who, in the contracts covered
by this Directive, is acting for purposes which are not related to his trade,
business or profession; [. . . ]
that has a different meaning with respect to the definition of consumer given in
the Council Directive 90/314/EEC, Art. 2.4:
[. . . ] “consumer” means the person who takes or agrees to take the package
(‘the principal contractor’), or any person on whose behalf the principal con-
tractor agrees to purchase the package (‘the other beneficiaries’) or any person
to whom the principal contractor or any of the other beneficiaries transfers
the package (‘the transferee’) [. . . ]
The LTS should be able to represent both the more specific dimension related
to the definitions in EUDs and the more abstract one which results from the
doctrinal interpretation of European law. The LTS allows inserting the text
paragraphs where umeanings are defined. However, to gain better understanding
of legal concepts, it is often required to consider a broader fragment. For example,
in the case of consumer the definition is not enough, and it is necessary to
collect multiple paragraphs where consumer protection norms are presented and
discussed.
2.2 Normative Change
Another big open issue to cope with in building tools for describing legal frame-
works both at the European and at the national level is the normative change [12].
One major problem, well-known in the literature, is the update of non-monotonic
ontologies and knowledge bases [4]. In other words, not necessarily ontologies and
knowledge bases have a structure constant through time (e.g., see [16]): concepts
and relations present in the ontology can become obsolete as new concepts and
relations are added. This is indeed the case of legal frameworks, that are contin-
uously modified as new laws can modify paragraphs of old ones.
We can have two types of normative change: explicit change and implicit
change. In the first case the new norm explicitly states the abrogation of a
specific paragraph of an old law (for details on this line of investigation, please
refer to [6]). Alternatively, the newer law can state a concept in contradiction to
previous laws, but without mentioning them explicitly. In this case the concept
stated by the new law becomes the current one; also, the parts of the old laws
affected by changes (no longer updated) become obsolete.
3 From LTS 1.0 to LTS 2.0
In this Section we first summarize the functionalities of the existing LTS [2], and
then we explain how it has been extended to cope with the new requirements
described in the previous Section.
3.1 LTS 1.0
The main assumptions of our methodology come from studies in comparative law
[13] and ontologies engineering [10]. Terms –lexical entries for legal information–,
and concepts must be distinguished; for this purpose we use lightweight ontolo-
gies, i.e. simple taxonomic structures of primitive or composite terms together
with associated definitions. They are hardly axiomatized as the intended mean-
ing of the terms used by the community is more or less known in advance by
all members, and the ontology can be limited to those structural relationships
among terms that are considered as relevant.
We distinguish the ontology implicitly defined by EUD, the EU level, from
the various national ontologies. Each one of these “particular” ontologies belongs
to the national level : i.e., each national legislation refers to a distinct national
legal ontology. We do not assume that the transposition of an EUD automatically
Term-Ita-A Term-Ger-A
Ger-5
EU-1 Ger-3
Ita-2
Ita-4
Fig. 1. Relationship between ontologies and terms. The thick arcs represent the inter-
ontology “association” link.
introduces in a national ontology the same concepts that are present at the EU
level.
Corresponding concepts at the EU level and at the national level can be
denoted by different terms in the same national language.
A standard way to properly manage large multilingual lexical databases is
to make a clear distinction among terms and their interlingual acceptions (or
axies) [15].
In the LTS project to properly manage terminological and conceptual mis-
alignment, we distinguish the notion of legal term from the notion of legal concept
and we build a systematic classification based on this distinction. The basic idea
in our system is that the conceptual backbone consists in a taxonomy of con-
cepts (ontology) to which the terms can refer in order to express their meaning.
One of the main points to keep in mind is that we do not assume the exis-
tence of a single taxonomy covering all languages. In fact, the different national
systems may organize the concepts in different ways. For instance, the term con-
tract corresponds to different concepts in common law and civil law, where it
has the meaning of bargain and agreement, respectively [14]. In most complex
instances, there are no homologous between terms-concepts such as frutto civile
(legal fruit) and income, but respectively civil law and common law systems can
achieve functionally similar operational rules thanks to the functioning of the
entire taxonomy of national legal concepts [9]. Consequently, the LTS includes
different ontologies, one for each involved national language plus one for the
language of EU documents. Each language-specific ontology is related via a set
of association links to the EU concepts, as shown in Fig. 1.
Although this picture is conform to intuition, in the basic LTS it has been
implemented by taking two issues into account. First, it must be observed that
the various national ontologies have a reference language. This is not the case
for the EU ontology. For instance, a given term in English could refer either to
a concept in the UK ontology or to a concept in the EU ontology. In the first
case, the term is used for referring to a concept in the national UK legal system,
Cancellation Conclusione del contratto
Consumer protection Difesa del consumatore
Termination Diritto di recesso
Withdrawal Risoluzione
Recesso
EU-1
Ita-1 Ita-2
Eng-1 purpose
EU-2 purpose purpose
Eng-2 Ita-3 Ita-4
purpose
concerns concerns
is-a is-a
Eng-3 Eng-4 Ita-5 Ita-6
Fig. 2. An example of interconnections among terms.
whilst in the second one, it is used to refer to a concept used in the European
directives. This is one of the main advantages of LTS. For example klar und
verständlich could refer both to concept Ger-379 (a concept in the German
Ontology) and to concept EU-882 (a concept in the European ontology). This is
the LTS solution for facing the possibility of a partial correspondence between
the meaning of a term in the national system and the meaning of the same term in
the translation of a EU directive. This feature enables the LTS to be more precise
about what “translation” means. It makes available a way for asserting that two
terms are the translation of each other, but just in case those terms have been
used in the translation of an EU directive: within LTS, we can talk about direct
EU-to-national translations of terms, and about implicit national-to-national
translations of terms. In other words, we distinguish between explicit and implicit
associations among concepts belonging to different levels. The former ones are
direct links that are explicitly used by legal experts to mark a relation between
concepts. The latter ones are indirect links: if we start from a concept at a given
national level, by following a direct link we reach another concept at European
level. Then, we will be able to see how that concept is mapped onto further
concepts at the various national levels.
The situation enforced in LTS is depicted in Fig. 1, where it is represented
that the Italian term Term-Ita-A and the German term Term-Ger-A have been
used as corresponding terms in the translation of an EU directive, as shown by
the fact that both of them refer to the same EU-concept EU-1. In the Italian
legal system, Term-Ita-A has the meaning Ita-2. In the German legal system,
Term-Ger-A has the meaning Ger-3. The EU translations of the directive is
correct insofar no terms exist in Italian and German that characterize precisely
the concept EU-1 in the two languages (i.e., the “associated” concepts Ita-4 and
Ger-5 have no corresponding legal terms). A practical example of such a situa-
EU-50
Consumer
INTERPRETED_AS
INTERPRETED_AS
EU-25 EU-28
Description: Description:
... ...
References: References:
93/13/EEC, Art. 1 02/65/EC, Art. 2
Fig. 3. Umeanings Eu-25 and Eu-28 are interpreted by the more abstract umeaning
Eu-50, the link between Eu-50 and the term “consumer” is implicit.
tion is reported in Fig. 2, where we can see that the ontologies include different
types of arcs. Beyond the usual is-a (linking a category to its supercategory),
there are also the arcs purpose, which relate a concept to the legal principle mo-
tivating it, and concerns, which refer to a general relatedness. The dotted arcs
represent the reference from terms to concepts. Some terms have links both to a
National ontology and to the EU Ontology (in particular, withdrawal vs. recesso
and difesa del consumatore vs. consumer protection).
The last item above is especially relevant: note that this configuration of
arcs specifies that: 1) withdrawal and recesso have been used as equivalent terms
(concept EU-2) in some European Directives (e.g., Directive 90/314/EEC). 2) In
that context, the term involved an act having as purpose some kind of protection
of the consumer. 3) The terms used for referring to the latter are consumer
protection in English and difesa del consumatore in Italian. 4) In the British
legal system, however, not all withdrawals have this goal, but only a subtype
of them, to which the code refers to as cancellation (concept Eng-3). 5) In the
Italian legal system, the term diritto di recesso is ambiguous, since it can be used
with reference either to something concerning the risoluzione (concept Ita-4),
or to something concerning the recesso proper (concept Ita-3).
3.2 Enhancing LTS with interpretation and abstraction
As described in Section 2.1, different pieces of legislations can bear different
definitions of terms. Having different detailed definitions is important during the
interpretation of very sectorial legal cases, but for the general case it is important
to have a view that abstracts from the peculiarities of specific domains.
In order to solve this problem we introduced a new kind of ontologic relation
called INTERPRETED AS : it is a non transitive relation where the more general
umeaning, that we will call group leader represents the abstracted concept that
groups the meaning of a number of more specific umeanings, that are the sectorial
umeanings defined in the individual EUDs or national laws (see Fig. 3).
We have also introduced a number of constraints and integrity checks to
ensure that the semantics of the grouping concept is respected and to improve
the usability of the system: i ) each umeaning can belong to a single group; ii ) a
group leader cannot exist without group members; iii ) when the user searches
into the umeaning database, more specific umeanings are excluded from the
results unless the user explicitly asks to show them, i.e. only the group leaders
are shown in the results. The need to contextualize concepts to the EUDs defining
them leads to the need of more complex instruments to deal with the language
of the norms. An umeaning is defined by the legal texts themselves; this makes
clear that the creation of umeaning is a quite long task, because it requires from
the user searching and reading a very large number of documents.
In order to ease this process, we developed a database that contains the
full versions of the desired EUDs and national laws. In this way, the user can
carry out his task according to the following workflow. 1 ) The user creates a
new umeaning linked with the term he wants to define; 2 ) He selects relevant
citation from legal text; consequently, the browser is redirected to a search page
and the main term attached to the umeaning is used as the default query; 3 )
After choosing one of the search results, the full text of the legal document is
displayed, with the search terms highlighted; 4 ) Finally, the user selects the text
that will go in the citation with the mouse and confirms the insertion in the
references database. Lastly, when the user searches for a term in the documents
database, the search is not performed upon the exact words, rather with their
roots, so for instance when performing the search on the term “contracts” also
documents containing only “contract” will be found, this seems to enhance the
information retrieval performance as shown in [11].
3.3 LTS with normative change
When a new normative is approved and enacted it can define a number of new
umeanings; moreover it can happen that the same law can change a number
of old umeanings defined by old laws. In particular, these old umeaning can
become obsolete and no longer valid. We are aware of the difficulties concerning
the modelling of the time in artificial intelligence and in formal ontology too5 .
Anyway, in LTS we adopted a naive solution in order to manage the simpler
situation concerning t In the LTS it was necessary to delete all old umeanings,
causing the loss of all historic informations from the database, informations that
are quite valuable to better understand the evolving of the normative. This
problem was resolved by using the same solution adopted for the interpretation
and abstraction of the norms (Section 3.2), i.e. empowering LTS with a new
ontological relation called REPLACED BY.
When the paragraph of an EUD defining an umeaning has been modified by a
new EUD, the new one defines a new umeaning that will replace the old umean-
5
E.g. see [3] for a general survey and [12] for normative systems
REPLACED_BY
Eu38: consumer Eu50: consumer
Date: september, 23 2002
IS_A IS_A
Eu38: final consumer
Explicitly inserted relations
Automatically generated relations
Fig. 4. An example of use of the REPLACED BY relation.
ing in the ontology. There will be a relation of type REPLACED BY between
the two umeanings, where the child umeaning is replaced by the more general
umeaning. Also in this case the new ontological relation has some peculiar char-
acteristics that distinguishes it from the usual ontological relations (Figure 4):
i ) a REPLACED BY relation brings with it a new data field not present in the
other relations: the substitution date; ii ) when the user performs a search in the
umeanings database the replaced ones will not be shown, unless the user asks
for a certain past date, thus obtaining a snapshot of the legal ontology that was
valid in that particular moment; iii ) when a new umeaning replaces an old one
all the ontological relations where the old umeaning appeared are automatically
copied in the new umeaning. If some of them are no longer valid with the new
umeaning, manual intervention from the user is required.
4 Conclusions
In this paper we discuss some features that have recently been introduced in
the LTS, a tool for building multilingual conceptual dictionaries for the EU law.
The tool is based on lightweight ontologies to emphasize the distinction between
concepts and terms. Different ontologies are built at the EU level and for each
national language, to deal with polysemy and terminological and conceptual
misalignment.
The present work illustrates how to distinguish between concepts as they are
defined in the text of the directives and the concepts representing the doctrinal
interpretation of the terms. Moreover, we point out how to deal with normative
change by introducing a temporal dimension in ontologies.
Future work will involve exploring how to extend the LTS ontology, with
special focus on the issue of populating it at the various levels by semi-automatic
approaches [7].
References
1. G. Ajani, G. Boella, L. Lesmo, M. Martin, A. Mazzei, and P. Rossi. A develop-
ment tool for multilingual ontology-based conceptual dictionaries. In Proc. of 5th
International Conference on Language Resources and Evaluation, LREC06, pages
1–6, Genoa, 2006.
2. G. Ajani, G. Boella, L. Lesmo, A. Mazzei, and P. Rossi. Terminological and onto-
logical analysis of european directives: multilinguism in law. In 11th Internation
Conference o Arificial Intelligence and Law (ICAIL), pages 43–48, 2007.
3. J. F. Allen. Towards a general theory of action and time. Artificial Intelligence,
23(2):123–154, 1984.
4. M. Cadoli and F. M. Donini. A Survey on Knowledge Compilation. AI Commu-
nications, 10(3–4):137–150, 1997.
5. P. Casanovas, N. Casellas, C. Tempich, D. Vrandecic, and R. Benjamins. OPJK
modeling methodology. In Proceedings of the ICAIL Workshop: LOAIT 2005, 2005.
6. M. Cherubini, G. Giardiello, S. Marchi, S. Montemagni, P. Spinosa, and G. Venturi.
NLP-based metadata annotation of textual amendments. In Proc. of WORKSHOP
ON LEGISLATIVE XML 2008, JURIX 2008, 2008.
7. P. Cimiano. Ontology Learning and Population from Text: Algorithms, Evaluation
and Applications. Springer, 2006.
8. S. Després and S. Szulman. Merging of legal micro-ontologies from european di-
rectives. Journal of Artificial Intelligence and Law, February 2007.
9. M. Graziadei. Tuttifrutti. In P. Birks and A. Pretto, editors, Themes in Compar-
ative Law, pages –. Oxford University Press, 2004.
10. M. Klein. Combining and relating ontologies: an analysis of problems and solutions.
In Workshop on Ontologies and Information Sharing, IJCAI’01, Seattle, USA,
2001.
11. R. J. Krovetz. Word sense disambiguation for large text databases. PhD thesis,
University of Massachusetts, 1995.
12. M. Palmirani and R. Brighi. Time model for managing the dynamic of normative
system. Electronic Government, pages 207–218, 2006.
13. P. Rossi and C. Vogel. Terms and concepts; towards a syllabus for european private
law. European Review of Private Law (ERPL), 12(2):293–300, 2004.
14. R. Sacco. Contract. European Review of Private Law, 2:237–240, 1999.
15. G. Sérasset. Interlingual lexical organization for multilingual lexical databases in
NADIA. In Proc. COLING94, pages 278–282, 1994.
16. The Gene Ontology Consortium. Gene Ontology: tool for the unification of biology.
Nature Genetics http: // genetics. nature. com , 25:25–29, 2000.
17. Daniela Tiscornia. The Lois Project: Lexical Ontologies for Legal Information.
In Proceedings of the V Legislative XML Workshop. European Press Academic
Publishing, 2007.
18. P. Vossen, W. Peters, and J. Gonzalo. Towards a universal index of meaning. In
Proc. ACL-99 Siglex Workshop, 1999.