In this paper we provide a probabilistic interpretation for typed feature structures very similar to those used by Pollard nd Sag. We begin with a version of the interpretation which lacks a treatment of re-entrant feature struc- tures, then provide an extended interpre- tation which allows them. We sketch al- gorithms allowing the numerical parameters of our probabilistic interpretations of HPSG to be estimated from corpora.

Establishing a clean relationship between a robot’s spatial model and natural language components is a non-trivial task, but is key to designing verbally controlled, navigating service robots. In this paper we examine the issues involved in the development of dialogue controlled navigating robots. In particular, we treat our robots as so-called Shared Control Systems, where robot and user cooperate to achieve a shared goal. We begin by characterising four categories of Shared Control Problems that affect verbally controlled navigating robots. Producing solutions to these problems requires a clear methodology in the linking of ’common-sense ’ representations of space used by the robots, and the language interface. To this end, we present the SharC Cognitive Control Architecture as a general purpose, agent-based dialogue control system that provides a suitable framework for relating spatial information to natural language communication. To illustrate our approach, we focus in particular on natural language understanding, and show how natural language utterances may be mapped to formally modelled spatial concepts, thus helping to overcome problems in shared control.

In this paper, we present a definition of unification of weighted feature structures designed to deal with constraint relaxation. The application of phrase structure rules in a unification-based Natural Language Processing system is adapted such that inconsistent values do not lead to failure, but are penalised. These penalties are based on the signature and the shape of the feature structures, and thus realise an elegant and general approach to relaxation.

Recent research has shown that unification grammars can be adapted to incorporate statistical information, thus preserving the processing benefits of stochastic context-free grammars while offering an efficient mechanism for handling dependencies. While complexity studies show that a probabilistic unification grammar achieves an appropriately lower entropy estimate than an equivalent PCFG, the problem of parameter estimation prevents results from reflecting the empirical distribution. This paper describes how a PUG can be implemented as a Prolog DCG annotated with weights, and how the weights can be interpretted to give accurate entropy estimates. An algorithm for learning correct weights is provided, along with results from some complexity analyses. Introduction Probabilistic context-free grammars are a popular formalism for modeling language because they preserve the simple CFG framework for characterising syntactic structure while incorporating the predictive benefits of statistica...

Résumé: La Prothèse Vocale Intelligente est un système de communication qui reconstitue le sens — supposé — d’une séquence peu structurée d’icônes ou de symboles, et l’exprime par des phrases en langue naturelle (français). Elle a été développée pour l’usage de personnes ne possédant pas la maîtrise du langage oral, et en particulier incapables de s’exprimer en suivant les règles d’une grammaire complexe comme celle de la langue. Nous décrivons ici la construction d’un dictionnaire sémantique de symboles simple et pertinent à partir des corpus de communication icônique relevés auprès d’enfants Infirmes Moteurs Cérébraux (IMC). Nous expliquerons ensuite le mécanisme d’analyse sémantique ascendante qui permet, en déterminant les dépendances entre symboles, de trouver le sens des messages de l’utilisateur. À partir du résultat de cette analyse, un module de transfert lexical choisit les mots français les mieux adaptés pour l’exprimer, et construit un réseau sémantique linguistique. Celui-ci est ensuite hiérarchisé, grâce à une Grammaire d’Arbres Adjoints (TAG) lexicalisée, en arbres syntaxiques de phrases françaises. Enfin, nous décrirons l’interface d’accès paramétrable qui a été définie pour ce système.

this paper we discuss how we apply predictions from our plan-based discourse processor discussed in (Rose et al., 1995) to the problem of disambiguation. The work we report here has been done in the context of the Enthusiast Spanish to English translation system. The Enthusiast system is part of the JANUS Speech-to-Speech translation system (Woszcyna et al., 1993; Woszcyna et al., 1994; Suhm et al., 1994; Levin et al., 1995). We discuss and evaluate two different methods for combining context based predictions with non context based predictions, namely a genetic programming approach and a neural network approach. We demonstrate the advantage of incorporating context-based predictions over the purely non context-based approach discussed in (Lavie, 1995). The results presented here show a significant improvement over our previous results reported in (Levin et al., 1995). Introduction In this paper we discuss how we apply predictions from our plan-based discourse processor discussed in (Ros e et al., 1995) to the problem of disambiguation. The work reported here had been carried out in the context of the Enthusiast Spanish to English translation system (Woszcyna et al., 1993; Woszcyna et al., 1994; Suhm et al., 1994; Levin et al., 1995). The Enthusiast System is part of the JANUS Speech-to-Speech translation system Ambiguity is a major problem in a large scale machine translation system such as Enthusiast. This is because the parsing grammar must be large in order to cover the wide range of constructions which speakers use. Additionally, the flexibility of the GLR* skipping parser (Lavie, 1995) we use magnifies the problem. In this paper we Computational Linguistics Program, Philosophy, Carnegie Mellon University, 5000 Forbes Ave., Pittsburgh PA, 15213, cprose@cs.cmu.edu

This report describes research done in the context of a subproject of the HPCN project IMPACT. The IMPACT project is headed by the ING bank and is founded by the organization for High Performance Computing and Networking (HPCN). The aim of the specific subproject, in the context of which this report has been written, is to develop (techniques for) natural language interfaces to information resources, focusing on the use of high-performance computers to achieve acceptable response times. This report is part of the "Parallel Parsing I" research topic.

This document summarizes the major work on discourse processing in the Enthusiast Spanish-English speech translation project. Traditionally machine translation systems have processed sentences in isolation. Recently, there has been an interest in making use of discourse information in machine translation. In [15, 20], researchers at ATR advocate an illocutionary act based machine translation, arguing that equivalent sentence forms do not necessarily carry the same illocutionary forces across languages. In addition to speech act recognition, discourse information can be used for ellipsis and anaphora resolution to provide more accurate translation. Discourse processing has been shown to be useful in other natural language processing applications as well. For example, [45] reports that a discourse state can be used to select an appropriate language model for speech recognition and parsing to reduce the perplexity and improve recognition and parsing accuracy. Our work on discourse processing has been done in four major directions:

In this paper we provide a probabilistic interpretation for typed feature structures very similar to those used by Pollard and Sag. We begin with a version of the interpretation which lacks a treatment of reentrant feature structures, then provide an extended interpretation which allows them. We sketch algorithms allowing the numerical parameters of our probabilistic interpretations of HPSG to be estimated from corpora. 1