Tree-adjoining grammars (TAG) have been proposed as a formalism for generation based on the intuition that the extended domain of syntactic locality that TAGs provide should aid in localizing semantic dependencies as well, in turn serving as an aid to generation from semantic representations. We demonstrate that this intuition can be made concrete by using the formalism of synchronous tree-adjoining grammars. The use of synchronous TAGs for generation provides solutions to several problems with previous approaches to TAG generation. Furthermore, the semantic monotonicity requirement previously advocated for generation gram- mars as a computational aid is seen to be an inherent property of synchronous TAGs.

Abstract not found

We present an algorithm for generating strings from logical form encodings that improves upon previous algorithms in that it places fewer restrictions on the class of grammars to which it is applicable. In particular, unlike an Earley deduction generator (Shieber, 1988), it allows use of semantically nonmonotonic grammars, yet unlike topdown methods, it also permits left-recursion. The enabling design feature of the algorithm is its implicit traversal of the analysis tree for the string being generated in a semantic-head-driven fashion.

this paper, I review the problem, and attempt to highlight certain salient aspects of it that have been lost in the pursuing of solutions, in order to reconcile the apparently contradictory claims of the problem's intractability and its resolution. 2 Review of Natural-Language Generation

In this paper I will discuss the properties of a tactical generation approach that has become popular recently: head-driven bottom-up generation. It is assumed that bidirectional grammars written in some unificationor logic-based formalism define relations between strings and some representation, usually called logical form. The task for a generator is to generate for a given logical form the strings that are related to this logical form by the grammar. In the paper it will be shown that the `early' approaches to this conceivement of the generation problem such as [21], [31] and [7] are not entirely satisfactory for general purposes. Furthermore I will define a simple bottom-up generator, called BUG1 for reference, as prototypical for the head-driven bottom-up approach as defended by for example [30, 5, 23, 24]. I will argue that head-driven bottom-up generation is to be preferred because the order of processing is directed by the input logical form and the information available in le...

In this paper we discuss some requirements on natural language interfaces that are needed for supporting connected dialogue, and propose methods for meeting these requirements in the context of a system that has been developed at our department over the last couple of years. On the basis of data collected in Wizard-of-Oz simulations we argue that the dynamic discourse representation needed for a given application can be structured in terms of a tree of dialogue objects (moves, initiativeresponse units) and a score board, i.e. a list of parameters each of which keeps some information relevant to the system's interpretation and generation tasks. We also argue that the domain knowledge needed for supporting reference resolution can, and should be obtained from data of this kind. 1

Abstract. We present a novel ensemble of six methods for improving the efficiency of chart realization. The methods are couched in the framework of Combinatory Categorial Grammar (CCG), but we conjecture that they can be adapted to related grammatical frameworks as well. The ensemble includes two new methods introduced here— feature-based licensing and instantiation of edges, and caching of category combinations—in addition to four previously introduced methods— index filtering, LF chunking, edge pruning based on n-gram scores, and anytime search. We compare the relative contributions of each method using two test grammars, and show that the methods work best in combination. Our evaluation also indicates that despite the exponential worstcase complexity of the basic algorithm, the methods together can constrain the realization problem sufficiently to meet the interactive needs of natural language dialogue systems. 1

this paper is that neither has been implemented." ([Vaughan and McDonald, 1986], page 95). Although Meteer [1990] gives a detail description of the relationship between text structure and revision it is unclear how the proposed model could contribute to the choice problem of paraphrases (see section 5.2). How- ever, from the approach described above and from the system described in [Meteer, 1990] we can draw the following conclusions. Only the generatoFs input is marked. If the generator encounters alternative realizations the revision component is asked to make the decision. However, to be able to do this it needs detailed knowledge about the grammar. Therefore grammatical knowledge has to be duplicated. The linguistic realization component used in [Meteer, 1990] is MUMBLE-86 [McDonald, 1986]. The text structural representation level must completely specify the infor- mation to be expressed by the utterance. The mapping has to ensure that all the necessary linguistic information is present. Mumblers procedural grammar is used only for generation purposes. Therefore it is without reach for the revision model to take into account relevant sources of ambiguities

Lexica! Grammars are a class of unification g,'ammars which share a fixed rule component, for which lhere exists a simple left-recursion elimination transformation. The parsing and generation programs ae seen as two dual non-left-recursive versions of the original grammar, and are implemented through a standard top-down Prolog interpreter. Formal criteria for termination are given as conditions on lexical enlrics: during parsing as well as during generution the processing of a lexical entry consumes some amount of a guide; the guide used for parsing is a list of words remaining to be analyzed, while the guide for genration is a list of the semantics of constituents waiting to be generated.

The paper describes the application of Explanationbased Learning for efficient processing of constraintbased grammars. The idea is to generalize the derivations of training instances created by normal parsing automatically and to use these generalized derivations (called templates) during the run-time mode of the system. In case a template can be instantiated for a new input, no further grammatical analysis is necessary. The approach is not restricted to sentential level but can be applied also to arbitrary phrases. Therefore, the EBL method can be interleaved straightforwardly with normal processing to get back flexibility that otherwise would be lost. 1 Introduction In the paradigm of feature-based grammar formalisms and theories, one line of research has been oriented towards developing high level languages which are based exclusively on constraint specifications and which allow grammar writers to specify different levels of linguistic abstraction in one homogeneous data structure...