Invisible meaning relations for representing near equivalents

Abstract

One of the key design principles of the Ekilex dictionary writing system is its symmetrical many-to-many relationship between word and meaning. Ekilex is currently being used for creating the EKI Combined Dictionary (CombiDic), with a primary goal of increasing coverage of languages beyond Estonian. This paper discusses the pilot project of integrating English, which began with generating a list of candidate equivalents for post-editing. The primary focus of the paper is on how near equivalents (narrower, wider, approximate) are represented in the symmetrical data model. Since meanings are language-independent entities in such a model, and equivalence is essentially about similarity of meanings, the near equivalents are represented using relations between meanings. To the dictionary user, the relations remain invisible and are only queried to retrieve target-language words for display. Transitioning from the traditional flowing text in the target language field to this more structured approach significantly affects the work process. We examine the advantages and disadvantages of this change in the paper.