Translation Algorithms by Means of Language Intersection

Synchronous rewriting grammars have been successfully exploited as translation models in machine translation applications. In this article we consider the problem of the design of translation algorithms based on synchronous context-free grammars. We revisit a methodology for the design of parsing algorithms based on the idea of language intersection, that has fully been developed in the parsing community, and show how to apply it in the design of translation algorithms. We argue that the intersection methodology above can also be viewed as a framework for the comparison of translation algorithms and for the formal analysis of their properties. On this line, we observe how superficially different translation algorithms that have been proposed in the literature can be viewed as special cases of application of the intersection methodology. We also use our framework to reformulate and improve some results already presented in the literature.