Towards a Development Methodology for Managing Linguistic Knowledge Bases

We are on the way of defining a methodology aimed to create software tools supporting linguistic knowledge bases. One of our main concerns is to formally represent knowledge using a sound software engineering approach. In this setting, we first consider the linguistic concepts found in a multilingual dictionary, as vocabulary, meanings, semantic categories, semantic relationships, and (treeshaped) taxonomy. Next, we start on ontologies, considering concepts as syntactic categories, orthography, phonology, syntactic features, lexical semantics and relations, and so on. We have represented in the conceptual levels these concepts, by using the well known entity-relationship model. In addition, we have applied the design cycle of databases in order to also fulfill the logical and physical models for representing the linguistic concepts at each development stage. In addition, we have developed both authoring and querying tools for both stages, and a migration procedure for interfacing them. Taking into account that the different existing linguistic knowledge bases have been built without following any formal methodology, our approach adds a way of integrating linguistic resources with a common structure. The resulting framework is useful for several applications, including multilingual information retrieval, document classification, and language translation, and also for their exploitation in education.