An approach to abductive inference developed in the TACITUS project has resulted in a dramatic simplification of how the problem of interpreting texts is conceptualized. Its use in solving the local pragmatics problems of reference, compound nominals, syntactic ambiguity, and metonymy is described and illustrated. It also suggests an elegant and thorough integration of syntax, semantics, and pragmatics. 1

The problems of access -- retrieving linguistic structure from some mental grammar -- and disambiguation -- choosing among these structures to correctly parse ambiguous linguistic input -- are fundamental to language understanding. The literature abounds with psychological results on lexical access, the access of idioms, syntactic rule access, parsing preferences, syntactic disambiguation, and the processing of garden-path sentences. Unfortunately, it has been difficult to combine models which account for these results to build a general, uniform model of access and disambiguation at the lexical, idiomatic, and syntactic levels. For example psycholinguistic theories of lexical access and idiom access and parsing theories of syntactic rule access have almost no commonality in methodology or coverage of psycholinguistic data. This paper presents a single probabilistic algorithm which models both the access and disambiguation of linguistic knowledge. The algorithm is based on a parallel parser which ranks constructions for access, and interpretations for disambiguation, by their conditional probability. Low-ranked constructions and interpretations are pruned through beam-search; this pruning accounts, among other things, for the garden-path effect. I show that this motivated probabilistic treatment accounts for a wide variety of psycholinguistic results, arguing for a more uniform representation of linguistic knowledge and for the use of probabilisticallyenriched grammars and interpreters as models of human knowledge of and processing of language.

While it has been realized for quite some time within AI that abduction is a general model of explanation for a variety of tasks, there have been no empirical investigations into the practical feasibility of a general, logic-based abductive approach to explanation. In this paper we present extensive empirical results on applying a general abductive system, Accel, to moderately complex problems in plan recognition and diagnosis. In plan recognition, Accel has been tested on 50 short narrative texts, inferring characters ' plans from actions described in a text. In medical diagnosis, Accel has diagnosed 50 real-world patient cases involving brain damage due to stroke (previously addressed by set-covering methods). Accel also uses abduction to accomplish model-based diagnosis of logic circuits (a full adder) and continuous dynamic systems (a temperature controller and the water balance system of the human kidney). The results indicate that general purpose abduction is an e ective and e cient mechanism for solving problems in plan recognition and diagnosis. 1

This paper is a preliminary effort to provide such a framework. In Section 2 the IA framework is presented, in just enough detail to allow this paper to stand on its own. The interested reader should consult IA and the other cited papers for a deeper discussion of the framework. In Sections 3 and 4, two examples are given. In the first, the informational reading and the intentional reading are essentially the same. In the second, they are in conflict and the intentional reading wins out. Section 5 summarizes. 2 Background

Narrative understanding is a hard problem for artificial intelligence that requires deep semantic understanding of natural language and broad world knowledge. Early research in this area stalled due to the difficulty of knowledge engineering and a trend in the field towards robustness at the expense of depth. This work explores how a practical integration of more recent resources and theories for natural language understanding can perform deep semantic interpretation of narratives when guided by specific pragmatic constraints. It shows how cognitive models can provide pragmatic context for narrative understanding in terms of well-defined reasoning tasks, and how those tasks can be used to guide interpretation and evaluate understanding. This work presents an implemented system, EA NLU, which has been used to interpret narrative text input to cognitive modeling simulations. EA NLU integrates existing large-scale knowledge resources with a controlled grammar and a compositional semantic interpretation process to generate highly expressive logical representations of sentences. Delayed disambiguation and representations from dynamic logic are used to separate this compositional process from a query-driven discourse interpretation process that is guided by pragmatic concerns and uses world knowledge. By isolating explicit points of ambiguity and using limited evidential abduction, this query-driven process can automatically identify the disambiguation choices that entail relevant interpretations. This work shows how this approach maintains

In the paper \Interpretation as Abduction " (hereafter IA) Hobbs et al. (1992) present and elaborate the view that to interpret an utterance is to nd the best explanation of why it would be true. We may call this the \Informational Perspective " on discourse interpretation. The only thing to be explained is the information explicitly conveyed by

Objects Finally, no discussion of linguistic description could be complete without a discussion of the range of further abstract objects that grammar implicates, such as STATES, POSSIBILITIES, FACTS and PROPOSITIONS. But there is perhaps one notable difference between these abstractions and the individuals we have already considered. A specification or system may be equipped with an inventory of concrete individuals to describe, by listing the objects or places in its environment, the meaningfully different quantities it can reason about, and even the events in some history or plan. But to describe abstract individuals in interesting ways, a system needs a generative understanding of them, and needs to create representations of them by inference, on the fly. Such inference in turn requires a more sophisticated implementation. At the same time, the linguistic expressions that describe abstractions are more rarefied. This gives two reasons why knowledge representation does not approach abstractions with the same urgency as other individuals. Of course, abstract individuals remain quite intriguing, as we shall now discover.

This paper describes a new, psychologically-plausible model of human sentence interpretation, based on a new model of linguistic structure, Construction Grammar. This on-line, parallel, probabilistic interpreter accounts for a wide variety of psycholinguistic results on lexical access, idiom processing, parsing preferences, and studies of gap-filling and other valence ambiguities, including various frequency effects. We show that many of these results derive from the fundamental assumptions of Construction Grammar that lexical idioms, idioms, and syntactic structures are uniformly represented as grammatical constructions, and argue for the use of probabilistically-enriched grammars and interpreters as models of human knowledge of and processing of language. Submitted to Cognitive Science ii 1 Introduction In the last twenty years, the field of cognitive science has seen an explosion in the number of computational models of cognitive processing. This is particularly true in the mode...

A diverse set of intelligent activities, including natural language understanding and diagnosis, requires the ability to construct explanations for observed phenomena. In this paper, we view explanation as abduction, where an abductive explanation is a consistent set of assumptions which, together with background knowledge, logically entails a set of observations. We have successfully built a domain-independent system, Accel, in which knowledge about a variety of domains is uniformly encoded in rst-order Hornclause axioms. A general-purpose abduction algorithm, AAA, e ciently constructs explanations in the various domains by caching partial explanations to avoid redundant work. Empirical results show that caching of partial explanations can achieve more than an order of magnitude speedup in run time. We have applied our abductive system to two general tasks: plan recognition in text understanding, and diagnosis of medical diseases, logic circuits, and dynamic systems. The results indicate that Accel is a general-purpose system capable of plan recognition and diagnosis, yet e cient enough to be of practical utility.

The paper explains the absence of resultative secondary predication in Russian as arising from a con°ict of inferential interpretations. It formalises the framework necessary to express this proposal in terms of abductive reasoning with Poole systems in Gricean contexts. The con°ict is shown to arise for default rules regulating alternative realisation of verb-internally speci¯ed consequent states. The paper thus indicates that typological variation may be due not only to di®erent parameter values but to general inferential properties of the syntax-semantics mapping. The proposed theory also contradicts some widespread proposals that the absence of resultative secondary predication is due to the absence of some particular language feature. 1