Towards the Global Wordnet

The global wordnet is an extensive lexical-semantic
network that constitutes of synonymous sets
(synsets) linked with the semantic relations existing between them. The cross-lingual nature of the global wordnet is provided by the establishing of relations of equivalents between synsets that express the same
meaning in different languages. The global wordnet
offers not only the extensive data for the comparative analysis over lexical densities and levels of lexicalization
but furthermore presupposes the successful implementation
in different application areas such as cross-lingual
information and knowledge management, cross-lingual
content management and text data mining, cross-lingual
information extraction and retrieval, multilingual
summarization, machine translation, etc. Therefore the proper maintaining of the completeness and consistency of the global wordnet is an important prerequisite for any type of text processing to which it is intended.
The EuroWordNet (EWN) extended the Princeton
wordnet (PWN) with cross-lingual relations [Vossen, 1999], which were further adopted by BalkaNet (BWN) [Stamou, 2002]. The languages covered by the EWN are Czech, Dutch, Estonian, French, German, Italian, and Spanish, respectively, and those covered by the BWN are Bulgarian, Greek, Romanian, Serbian and Turkish. The equivalent synsets in different languages are linked to
the same Inter-Lingual Index (ILI) thus connecting
monolingual wordnets in a global lexical-semantic network.
The Inter-Lingual Index is based on the PWN (ILI is consecutively synchronized with the PWN versions), the synsets of which are considered as language independent
concepts. Thus a distinction between the language-
specific modules (English among them) and the language-independent module (the ILI repository) has to be focused. The ILI is considered as an unstructured list of meanings, where each ILI-record consists of a synset (if the language is not English, a proper translation or at least transliteration must be ensured), an English gloss specifying the meaning and a reference to its source.
Both EWN and BWN adopted the hierarchy of concepts and relations’ structure of the English wordnet as a model to be followed in the development of each language-specific wordnet. For the monolingual wordnets a strong rule is observed – strictly to preserve the structure of the PWN because via the ILI a proper cross-lingual navigation is ensured. It is natural, that some of the concepts stored
in ILI are not lexicalized in all languages and there are language specific concepts that might have no ILI
equivalent. In the first case, the empty synsets were created
(called non-lexicalized synsets) in the wordnets for the languages that do not lexicalize the respective concepts. The non-lexicalized synsets preserve the hierarchy and their purpose is to cover the proper cross-lingual
relations. Regarding the second case, the ILI is further extended both in EWN and BWN with some language specific concepts. The language specific concepts that are shared between Balkan languages are linked via a BILI (BalkaNet ILI) index [Tufis, 2004]. The initial set of common Balkan specific concepts consisted mainly of concepts reflecting the cultural specifics of the Balkans
(family relations, religious objects and practices,
traditional food, clothes, occupations, arts, important events, measures, etc).
There are four morpho-semantic relations included in PWN and mirrored in EWN and BWN, Be in state,
Derivative, Derived and Participle [Koeva, 2004]. Those
relations semantically linked synsets although they can actually be applied to the literals only (graphic and
compound lemmas). Consider the following examples:
Be in state is an asymmetric inverse intransitive relation that links derivationally and semantically related adjectives and nouns. The English synset {attractive:3, magnetic:5} with a definition ‘having the properties of a magnet; the ability to draw or pull’ is in a Be in state relation with the synset {magnetism:1, magnetic attraction:1, magnetic force:1} with a definition ‘attraction for iron; associated with electric currents as well as magnets; characterized
by fields of force’; also the synset {attractive:1}
with a definition ‘pleasing to the eye or mind especially through beauty or charm’ is in a Be in state relation with {attractiveness:2} denoting ‘a beauty that appeals to the senses’.
Derivative is an asymmetric inverse intransitive relation
between derivationally and semantically related noun and verb synsets. For example the English synset
{rouge:1, paint:3, blusher:2} with a definition ‘makeup consisting of a pink or red powder applied to the cheeks’ is in Derivative relation with two synsets: {rouge:1} with a meaning ‘redden by applying rouge to’ and {blush:1, crimson:1, flush:1, redden:1} denoting ‘turn red, as if in embarrassment or shame’.
Derived is an asymmetric inverse intransitive relation between derivationally and semantically related adjective and noun synsets. For example the synset {Cuban:1} with a definition ‘of or relating to or characteristic of Cuba or the people of Cuba’ is in a Derived relation with {Cuba:1, Republic of Cuba:1}.
Participle is an asymmetric inverse intransitive relation
between derivationally and semantically related an
adjective synset denoting result of an action or process and the verb synset denoting the respective action or process.
Consider {produced:1} with a definition ‘that is caused by’ which is in a Participle relation with {produce:3, bring about:4, give rise:1} denoting ‘cause to occur or exist’.
As can be seen by the examples, although the synsets are semantically linked, the actual derivational relations are established between particular literals. For the best
performance of the multilingual data base in different text processing tasks a specification of the derivational links must to be kept at the level of literal notes (LNotes).
There are systematic morpho-semantic differences
between English and Slavic languages – namely
derivational processes for building relative adjectives, gender pairs and diminutives. The Slavic languages
possess rich derivational morphology which has to be
involved into the strict one-to-one mapping with the ILI.
A vivid derivational process rely Slavic nouns with
respective relative adjectives with general meaning ‘of or related to the noun’. For example, the Bulgarian relative
adjective {стоманен:1} defined as ‘of or related to steel’ has the Serbian equivalent {čelični:1} with exactly the same definition. Actually in English this relation is
expressed by the respective nouns used with an adjectival
function (rarely at the derivational level, consider wooden↔wood, golden↔gold), thus the concepts exist in English and the mirror nodes have to be envisaged.
The gender pairing is systematic phenomenon in
Slavic languages that display binary morpho-semantic opposition: male↔female, and as a general rule there is no corresponding concept lexicalized in English. The derivation is applied mainly to nouns expressing
professional occupations. For example, Bulgarian synset {преподавател:2, учител:1, инструктор:1} and Serbian
synset {predavač:1} that correspond to the English
{teacher:1, instructor:1} with a definition: ‘a person whose
occupation is teaching’ have their female gender counterparts
{преподавателка, учителка, инструктурка} and {predavačica} with a feasible definition ‘a female
person whose occupation is teaching’. There are some exceptions where like in English one and the same word is used both for masculine and feminine in Bulgarian and Serbian, for example {президент:1} which corresponds to the English synset {president:3} with a definition: ‘the chief executive of a republic’, and as a tendency the
masculine noun can be used referring to females.
Diminutives are standard derivational class for expressing concepts that relate to small things. The diminutives display
a sort of morpho-semantic opposition: big ↔ small,
however sometimes they may express an emotional
attitude too. Thus the following cases can be found with diminutives: standard relation big ↔ small thing,
consider {стол:1} corresponding to English {chair:1} with a meaning ‘a seat for one person, with a support for the back’ and {столче} with an feasible meaning ‘a little seat for one person, with a support for the back’; small thing to which an emotional attitude is expressed. Also, Serbian synset {lutka:1} that corresponds to the English {doll:1, dolly:3} with a meaning ‘with a replica of a
person, used as a toy’ is related to {lutkica} which has both diminutive and hypocoristic meaning. There might be some occasional cases of the expression of that kind of concepts in English, {foal:1} with a definition: ‘a young horse’, {filly:1} with a definition: ‘a young female horse under the age of four’, but in general these concepts are expressed by phrases.
There are several possible approaches for covering
different lexicalization at different languages [Vitas & Krstev, 2005]:
- treat them as denoting specific concepts and define appropriate synsets;
- include them in the synset with the word they were derived from;
- omit their explicit mentioning, but rather let the flexion-derivation description encompass these
phenomena as well.
Treating morpho-semantic relations, relative adjectives,
gender pairs and diminutives, in Slavic languages as
relations that involve language specific concepts requires
an ILI addition for the languages where the concepts are presented (respectively lexical gaps in the rest). This
solution takes grounds from the following observations:
- relative adjectives, feminine gender pairs and
diminutives denote an unique concept;
- relative adjectives, feminine gender pairs and
diminutives are lexicalized with a single word
in Bulgarian, Serbian, Czech and other Slavonic
languages;
- relative adjectives, feminine gender pairs and
diminutives in most of the cases belong to different word class comparing to the word from which they are derived (there are some exceptions, like
diminutives that are derived from neuter nouns in Bulgarian) .
Moreover, as with the other morpho-semantic relations, a special attribute assigned at the LNotes must provide information for one-to-one derivational relations.
Although PWN’s coverage does not compare yet with new wordnets, the latter are continuously extended and improved so that a balanced global multilingual wordnet
is foreseen, thus the task of the proper encoding of
different level of lexicalization if different languages is in a great importance regarding the Natural Language Processing.
References
[Koeva at al., 2004] S. Koeva, T. Tinchev and S.
Mihov Bulgarian Wordnet-Structure and Validation in:
Romanian Journal of Information Science and
Technology, Volume 7, No. 1-2, 2004: 61-78.
[Stamou, 2002] Stamou S., K. Oflazer, K. Pala, D. Totkov, D. Dutoit, M. Grigoriadou, BALKANET:
A Multilingual Semantic Network for the Balkan Languages, Proceedings of the International Wordnet
Conference, Mysore, India, 21-25 January 2002,
12-14.
[Tufis, 2004] D. Tufis, D. Cristea, S. Stamou BalkaNet: Aims, Methods, Results and Perspectives.
A General Overview in: Romanian Journal of Information Science and Technology, Volume 7, No. 1-2, 2004: 1-32.
[Vitas & Krstev, 2005] Duško Vitas, Cvetana Krstev (2005) Derivational Morphology in an E-Dictionary
of Serbian in Proceedings of 2nd Language &
Technology Conference, April 21-23, 2005, Poznań, Poland, ed. Zygmunt Vetulani, pp. 139-143,
Wydawnictwo Poznańskie Sp. z o.o., Poznań, 2005.
[Vossen, 1999] Vossen P. (ed.) EuroWordNet: a
multilingual database with lexical semantic networks for European Languages. Kluwer Academic
Publishers, Dordrecht. 1999.