Generally, the portability of the NLI is defined as the possibility for these interfaces to be adapted in order to be used again in different data processing environments. these environments depend on many levels: the hardware system level, the data management system level (DBMS, Operating system, ...), the knowledge-domain level (geographic, medicine, ...) and the natural language level (Arabic, French, English, ...).
The majority of proposed researches for the design of NLI use an intermediate representation that promotes system and knowledge-domain portability. On the contrary, the linguistic portability had been neglected. Starting from the previously-mentioned definition of the portability, we will try to give a formal definition of the linguistic portability. This definition gives more precision and would serve as a basis for the quantification of this notion that remained fuzzy and subjective.
The basic idea of this reflection is to render the portability (i.e., linguistic or any other) to a rate of re-use of the interface components when one of the environment changes.
We will start with the presentation of a formalism proposed for the linguistic portability. Then we will present the main motivations for a semantic intermediate representation of queries and an object implementation. Next, we will deal with the semantic of the queries and their typologies. Finally, we will present the object model of our representation with its different links and classes illustrating the semantics of the query elements.