A mapping mechanism of NL sentences onto an SPN state machine for understanding purposes

Abstract

This paper presents a formal language (called GLOSSA) for the representation of natural language (NL) sentences. Specifically, GLOSSA is a context-free language devoted to describe different forms of a kernel (Agent→Action→Patient) of a NL sentence based on its components, which are agents, actions, and patients, and for different kind of connections between them. GLOSSA is also able to describe different kinds of connections between more than one kernels in the same NL sentence. The definition of this formal language is motivated by our future goal to convert NL sentences onto state machines. In order to support this transformation we propose GLOSSA, which offers a formal model for the representation of basic components of a NL sentence (agents, actions, patients). Thus, in this work we provide the formal definition of the GLOSSA language and also present some illustrative examples of its use.