The uniform tabular algorithm developed so far is new and exhibits several relevant novel ideas, e.g., a dynamic selection function for parsing and generation and the use of a uniform indexing mechanism for both tasks. It combines previous approaches for parsing and generation in one computational framework, and shows that parsing and generation can be realized uniformly and efficiently. Since we have considered parsing and generation in an isolated way, we are still in the main stream of grammatical processing.
However, we now present a novel method for grammatical processing, namely the use of items produced in one direction (e.g., parsing) directly in the other direction (e.g., generation). We will call this method item sharing between parsing and generation.
If one assumes that parsing and generation are performed in an isolated way, then such method seems to be an overhead. However, in the next chapter we will argue at length that a tight integration of parsing and generation is necessary in order to handle performance aspects like monitoring, revision or generation of paraphrases. On the basis of this discussion we present a set of new methods (in order of increasing complexity) that benefit from the uniform tabular algorithm in combination with the item sharing approach.
To my knowledge, the possibility of sharing items between parsing and generation has not been addressed in the literature so far. Under this perspective, our item sharing approach is a new dimension in the area of uniform processing of reversible grammars and demonstrates that the integration of parsing and generation can be achieved in an efficient way.