Computation And Language

The space and run-time requirements of broad coverage grammars appear for many applications unreasonably large in relation to the relative simplicity of the task at hand. On the other hand, handcrafted development of application-dependent grammars is in danger of duplicating work which is then difficult to re-use in other contexts of application. To overcome this problem, we present in this paper a procedure for the automatic extraction of application-tuned consistent subgrammars from proved large-scale generation grammars. The procedure has been implemented for large-scale systemic grammars and builds on the formal equivalence between systemic grammars and typed unification based grammars. Its evaluation for the generation of encyclopedia entries is described, and directions of future development, applicability, and extensions are discussed.

This paper presents a method for the automatic extraction of subgrammars to control and speeding-up natural language generation NLG. The method is based on explanation-based learning (EBL). The main advantage for the proposed new method for NLG is that the complexity of the grammatical decision making process during NLG can be vastly reduced, because the EBL method supports the adaption of a NLG system to a particular use of a language.

Studies of the contextual and linguistic factors that constrain discourse phenomena such as reference are coming to depend increasingly on annotated language corpora. In preparing the corpora, it is important to evaluate the reliability of the annotation, but methods for doing so have not been readily available. In this report, I present a method for computing reliability of coreference annotation. First I review a method for applying the information retrieval metrics of recall and precision to coreference annotation proposed by Marc Vilain and his collaborators. I show how this method makes it possible to construct contingency tables for computing Cohen's Kappa, a familiar reliability metric. By comparing recall and precision to reliability on the same data sets, I also show that recall and precision can be misleadingly high. Because Kappa factors out chance agreement among coders, it is a preferable measure for developing annotated corpora where no pre-existing target annotation exists.

Multiplicative weight-updating algorithms such as Winnow have been studied extensively in the COLT literature, but only recently have people started to use them in applications. In this paper, we apply a Winnow-based algorithm to a task in natural language: context-sensitive spelling correction. This is the task of fixing spelling errors that happen to result in valid words, such as substituting {\it to\/} for {\it too}, {\it casual\/} for {\it causal}, and so on. Previous approaches to this problem have been statistics-based; we compare Winnow to one of the more successful such approaches, which uses Bayesian classifiers. We find that: (1)~When the standard (heavily-pruned) set of features is used to describe problem instances, Winnow performs comparably to the Bayesian method; (2)~When the full (unpruned) set of features is used, Winnow is able to exploit the new features and convincingly outperform Bayes; and (3)~When a test set is encountered that is dissimilar to the training set, Winnow is better than Bayes at adapting to the unfamiliar test set, using a strategy we will present for combining learning on the training set with unsupervised learning on the (noisy) test set.

We describe an implemented system for robust domain-independent syntactic parsing of English, using a unification-based grammar of part-of-speech and punctuation labels coupled with a probabilistic LR parser. We present evaluations of the system's performance along several different dimensions; these enable us to assess the contribution that each individual part is making to the success of the system as a whole, and thus prioritise the effort to be devoted to its further enhancement. Currently, the system is able to parse around 80% of sentences in a substantial corpus of general text containing a number of distinct genres. On a random sample of 250 such sentences the system has a mean crossing bracket rate of 0.71 and recall and precision of 83% and 84% respectively when evaluated against manually-disambiguated analyses.

Although adequate models of human language for syntactic analysis and semantic interpretation are of at least context-free complexity, for applications such as speech processing in which speed is important finite-state models are often preferred. These requirements may be reconciled by using the more complex grammar to automatically derive a finite-state approximation which can then be used as a filter to guide speech recognition or to reject many hypotheses at an early stage of processing. A method is presented here for calculating such finite-state approximations from context-free grammars. It is essentially different from the algorithm introduced by Pereira and Wright (1991; 1996), is faster in some cases, and has the advantage of being open-ended and adaptable.

Currently, computational linguists and cognitive scientists working in the area of discourse and dialogue argue that their subjective judgments are reliable using several different statistics, none of which are easily interpretable or comparable to each other. Meanwhile, researchers in content analysis have already experienced the same difficulties and come up with a solution in the kappa statistic. We discuss what is wrong with reliability measures as they are currently used for discourse and dialogue work in computational linguistics and cognitive science, and argue that we would be better off as a field adopting techniques from content analysis.

This paper presents a corpus-based method to assign grammatical subject/object relations to ambiguous German constructs. It makes use of an unsupervised learning procedure to collect training and test data, and the back-off model to make assignment decisions.

There has recently been considerable interest in the use of lexically-based statistical techniques to resolve prepositional phrase attachments. To our knowledge, however, these investigations have only considered the problem of attaching the first PP, i.e., in a [V NP PP] configuration. In this paper, we consider one technique which has been successfully applied to this problem, backed-off estimation, and demonstrate how it can be extended to deal with the problem of multiple PP attachment. The multiple PP attachment introduces two related problems: sparser data (since multiple PPs are naturally rarer), and greater syntactic ambiguity (more attachment configurations which must be distinguished). We present and algorithm which solves this problem through re-use of the relatively rich data obtained from first PP training, in resolving subsequent PP attachments.

Deriving formal specifications from informal requirements is difficult since one has to take into account the disparate conceptual worlds of the application domain and of software development. To bridge the conceptual gap we propose controlled natural language as a textual view on formal specifications in logic. The specification language Attempto Controlled English (ACE) is a subset of natural language that can be accurately and efficiently processed by a computer, but is expressive enough to allow natural usage. The Attempto system translates specifications in ACE into discourse representation structures and into Prolog. The resulting knowledge base can be queried in ACE for verification, and it can be executed for simulation, prototyping and validation of the specification.