Bulgarian Pronouns in Universal Networking Language VELISLAVA STOYKOVA Institute for Bulgarian Language - IBL Bulgarian Academy of Sciences - BAS 52, Shipchensky proh. str. bl. 17, 1113 Sofia BULGARIA vstoykova@yahoo.com Abstract: The paper presents a computationally tractable application of Bulgarian pronouns representation using Universal Networking Language (UNL) formalism. It analyses grammar features of pronouns and offers a model of their formal representation based on incorporation of grammar, semantic and lexical knowledge by using standard UNL knowledge representation mechanisms. The entire application presents both morphological and syntactic knowledge and offers multilingual web-based application which can be further developed and elaborated. Key–Words: Semantic Networks, Knowledge Representation, Natural Language Processing, Machine Translation. 1 Introduction Pronouns exist in almost all European languages and are traditionally analysed as different part-of-speech with their own grammar features, semantics and spe- cific usage. The main function of pronouns is to sub- stitute another words or class of words, and with re- spect to that they have related grammar features. Pro- nouns’ grammar features are language-specific, how- ever, there are common view that pronouns play sig- nificant role in text mantaining its semantic relations and coherence. Also, some types of pronouns share semantics of deixis. Nevertheless, the main function of pronouns is to relate words in the sentence or in the text so that, they can be formally presented using various types of formalisms. Further, we are going to analyse formal interpretation of Bulgarian pronouns using semantic networks interpretation in UNL framework. 2 Bulgarian Pronouns The Bulgarian pronouns are traditionally described with respect to their semantics, grammar features, functions and usage. Generally, their semantics is connected to substitute, determine, relate, and agree both with other words in the sentence or within the whole text [Nicolova1986]. So, the grammar features of pronouns are con- nected to their semantics and functions and are highly dependent on agreement. In general, the pronouns have grammar features of person, number, and gen- der (for some also case, definiteness, etc.) These fea- tures can be successfuly used for formal representa- tions of pronouns with respect to computational ap- plications. Moreover, the existing formal semantic networks interpretation of possessive and reflexive- possessive pronouns using DATR language for lexical knowledge representation is based on the use of that features [Stoykova2010]. The Bulgarian pronouns are of following types: personal, reflexive, possessive, reflexive-possessive, demonstrative, relative, distributive, interrogative, in- definite, and negative. In our interpretation, we of- fer a systematic account for all types. Further, we are going to analyse and present semantic and grammar assumptions for related formal representation of Bul- garian pronouns. 3 The Universal Networking Lan- guage In the UNL approach, information conveyed by nat- ural language is represented as a hypergraph com- posed of a set of directed binary labeled links (re- ferred to as ”relations”) between nodes or hypernodes (the ”Universal Words”(UWs)), which stand for con- cepts. UWs can also be annotated with ”attributes” representing context information [Uchida et al.2005]. Universal Words (UWs) represent universal con- cepts and correspond to the nodes to be interlinked by ”relations” or modified by ”attributes” in a UNL graph. They can be associated to natural language open lexical categories (noun, verb, adjective and ad- verb). Additionally, UWs are organized in a hierarchy (the UNL Ontology), are defined in the UNL Knowl- Recent Advances in Computer Science ISBN: 978-960-474-311-7 142