The goal of the study is a design of a framework for formalized representation of lexical knowledge, as presented in bilingual dictionaries. Little research has been done on the possibilities of representation and storage of the knowledge acquired in the process of lexicographical analysis and used in the synthesis of dictionary entries. Separation of content from a particular form would allow for re-use of the data for several purposes (including NLP) and for flexible customization of dictionaries for different users.
In the first part, general abstract principles of representation of lexical knowledge are sought. The structure of different dictionary entries is analyzed. Modern technical approaches, which may contribute to an efficient representation of the knowledge, are summarized and a generic abstract model for its representation is defined in terms of objects and relations, together with a proposal for a modular implementation separating the language and dictionary specific components.
The second part demonstrates the use of the model for one particular task: a detailed description of a group of Norwegian nouns in contrast with their Czech equivalents. The nouns are analyzed and a possible representation of the knowledge is presented using the proposed generic model and task specific specifications.