Alex Lascarides

• BSc, MSc, PhD
• Professor (Full) at University of Edinburgh

Learning from demonstration is an effective method for human users to instruct desired robot behaviour. However, for most non-trivial tasks of practical interest, efficient learning from demonstration depends crucially on inductive bias in the chosen structure for rewards/costs and policies. We address the case where this inductive bias comes from...

Humans face many game problems that are too large for the whole game tree to be used in their deliberations about action, and very little is understood about how they cope in such scenarios. However, when a human player’s chosen strategy is conditioned on her limited perspective of how the game might progress (Degremont et al. 2016), then it should...

Effective human-robot interaction, such as in robot learning from human demonstration, requires the learning agent to be able to ground abstract concepts (such as those contained within instructions) in a corresponding high-dimensional sensory input stream from the world. Models such as deep neural networks, with high capacity through their large p...

This paper concerns the form-meaning mapping of communicative actions consisting of speech and improvised co-speech gestures. Based on the findings of previous cognitive and computational approaches, we advance a new theory in which this form-meaning mapping is analysed in a constraint-based grammar. Motivated by observations in naturally occurring...

As robots begin to cohabit with humans in semi-structured environments, the need arises to understand instructions involving rich variability---for instance, learning to ground symbols in the physical world. Realistically, this task must cope with small datasets consisting of a particular users' contextual assignment of meaning to terms. We present...

In this paper we present a comparative evaluation of various negotiation strategies within an online version of the game “Settlers of Catan”. The comparison is based on human subjects playing games against artificial game-playing agents (‘bots’) which implement different negotiation dialogue strategies, using a chat dialogue interface to negotiate...

Most models of rational action assume that all possible states and actions are pre-defined and that preferences change only when beliefs do. But several decision and game problems lack these features, calling for a dynamic model of preferences: preferences can change when unforeseen possibilities come to light or when there is no specifiable or mea...

We present a method for obtaining a useful symbolic model of persuasion in a complex game, where players' preferences over the outcomes of negotiations over resources are incomplete, uncertain, and dynamic. We focus on the problem of identifying the stage in the game where successfully persuading an agent to perform a particular action will have th...

We present an empirical framework for testing game strategies in The Settlers of Catan, a complex win-lose game that lacks any analytic solution. This framework provides the means to change different components of an autonomous agent's strategy, and to test them in suitably controlled ways via performance metrics in game simulations and via compari...

This paper describes a method that predicts which trades players execute during a win-lose game. Our method uses data collected from chat negotiations of the game The Settlers of Catan and exploits the conversation to construct dynamically a partial model of each player's preferences. This in turn yields equilibrium trading moves via principles fro...

We describe a dialogue model and an implemented annotation scheme for a pilot corpus of annotated online chats concerning bargaining negotiations in the game The Settlers of Catan. We will use this model and data to analyze how conversations proceed in the absence of strong forms of cooperativity, where agents have diverging motives. Here we concen...

This paper reports on an implementation of a multimodal grammar of speech and co-speech gesture within the LKB/PET grammar engi-neering environment. The implementation ex-tends the English Resource Grammar (ERG, Flickinger (2000)) with HPSG types and rules that capture the form of the linguistic signal, the form of the gestural signal and their rel...

In this article we make sdrt’s glue logic for computing logical form dynamic. This allows us to model a dialogue agent’s understanding of what the update of the semantic representation of the dialogue would be after his next contribution, including the effects of the rhetorical moves that he is contemplating performing next. This is a pre-requisite...

The use of hand gestures to point at objects and individuals, or to nav-igate through landmarks on a virtually created map is ubiquitous in face-to-face conversation. We take this observation as a starting point, and we demonstrate that deictic gestures can be analysed on a par with speech by using standard methods from constraint-based grammars su...

In this paper we propose a data intensive approach for inferring sentence-internal temporal relations. Temporal inference is relevant for practical NLP applications which either extract or synthesize temporal information (e.g., summarisation, question answering). Our method bypasses the need for manual coding by exploiting the presence of markers l...

We propose a method for modelling how dialogue moves influence and are influenced by the agents' preferences. We extract constraints on preferences and dependencies among them, even when they are expressed indirectly, by exploiting discourse structure. Our method relies on a study of 20 dialogues chosen at random from the Verbmobil corpus. We then...

Models of conversation that rely on a strong notion of cooperation don’t apply to strategic conversation — that is, to conversation where the agents’ motives don’t align, such as courtroom cross examination and political debate. We provide a game-theoretic framework that provides an analysis of both cooperative and strategic conversation. Our analy...

Résumé. Dans cet article, nous présentons une analyse à base de contraintes de la relation forme-sens des gestes déictiques et de leur signal de parole synchrone. En nous basant sur une étude empirique de corpus multimodaux, nous définissons quels énoncés multimodaux sont bien formés, et lesquels ne pourraient jamais produire le sens voulu dans la...

This paper addresses the form-meaning relation of multimodal communicative actions by means of a grammar that combines verbal input with hand gestures. Unlike speech, gesture signals are interpretable only through their semantic relation to the synchronous speech content. This relation serves to resolve the incomplete meaning that is revealed by ge...

Dialogue moves influence and are influenced by the agents’ preferences. We propose a method for modelling this interaction. We motivate and describe a recursive method for calculating the preferences that are expressed, sometimes indirectly, through the speech acts performed. These yield partial CP-nets, which provide a compact and efficient method...

The gestures that speakers use in tandem with speech include not only conventionalized actions with identifiable meanings (so-called narrow gloss gestures or emblems) but also productive iconic and deictic gestures whose form and meanings seem largely improvised in context. In this paper, we bridge the descriptive tradition with formal models of re...

Abstract We present an approach to the interpretation of non-sentential utterances like B’s utterance in the following mini-dialogue: A: “Who came,to the party?” B: “Peter.” Such utterances pose several puzzles: they convey ‘sentence-type’ messages (propositions, questions or request) while being of nonsentential form; and they are constrained both...

This paper provides a logically precise analysis of agreement and disputes in dialogue. The semantics distinguishes among the public commitments of each dialogue agent, including commitments to relational speech acts or rhetorical relations (e.g. Narration, Explanation and Correction ). Agreement is defined to be the shared entailments of the agent...

One way to construct semantic represen- tations in a robust manner is to enhance shallow language processors with seman- tic components. Here, we provide a model theory for a semantic formalism that is de- signed for this, namely Robust Minimal Recursion Semantics (RMRS). We show that RMRS supports a notion of entailment that allows it to form the...

This paper provides a logically precise analysis of grounding and disputes in dialogue. The semantics distinguishes among the public commitments of each dialogue participant, including commitments to relational speech acts or rhetorical relations (e.g., Narration, Explanation, Acknowledgement). Thus commitments to the illocutionary contribution of...

Being able to identify which rhetorical relations (e.g., contrast or explanation) hold between spans of text is important for many natural language processing applications. Using machine learning to obtain a classifier which can distinguish between different relations typically depends on the availability of manually labelled training data, which i...

We propose a dynamic semantics of questions in dialogue that tracks the public commitments of each dialogue agent, including commitments to issues raised by questions.

In this paper we define agreement in terms of shared public commitments, and implicit agreement is conditioned on the semantics of the relational speech acts (e.g., Narration, Ex- planation) that each agent performs. We pro- vide a consistent interpretation of disputes, and updating a logical form with the current utterance always involves extendin...

This paper motivates and describes a dynamic semantic theory of discourse interpre- tation called sdrt, which uses rhetorical relations to model the semantics/pragmatics interface. We describe the syntax and dynamic semantics of the language in which logical forms are represented, a separate but related language in which semantic underspeci- cation...

In face-to-face interaction, speakers make multimodal contributions that exploit both the linguis-tic resources of spoken language and the visual and spatial affordances of gesture. In this paper, we argue that, in formulating and understanding such multimodal contributions, interlocutors ap-ply the same principles of COHERENCE that characterize th...

We present a formal analysis of iconic coverbal gesture. Our model describes the incomplete meaning of gesture that's derivable from its form, and the prag- matic reasoning that yields a more spe- cific interpretation. Our formalism builds on established models of discourse in- terpretation to capture key insights from the descriptive literature on...

We describe a data-driven approach to building interpretable discourse structures for appointment scheduling dialogues. We represent discourse structures as headed trees and model them with probabilistic head-driven parsing techniques. We show that dialogue-based features regarding turn-taking and domain specific goals have a large positive impact...

We propose a method for automatically iden-tifying rhetorical relations. We use supervised machine learning but exploit cue phrases to au-tomatically extract and label training data. Our models draw on a variety of linguistic cues to distinguish between the relations. We show that these feature-rich models outperform the previ-ously suggested bigra...

We propose a novel method to predict the inter-paragraph discourse structure of text, i.e. to infer which paragraphs are related to each other and form larger segments on a higher level. Our method com-bines a clustering algorithm with a model of seg-ment "relatedness" acquired in a machine learning step. The model integrates information from a va-...

In this paper we propose a data intensive approach for inferring sentence-internal temporal relations, which relies on a simple probabilistic model and assumes no manual coding.

We present an approach to the interpretation of non-sentential utterances like B's utterance in the following mini-dialogue: A: "Who came to the party?" B: "Peter." Such utterances pose several puzzles: they convey `sentence -type' messages (propositions, questions or request) while being of nonsentential form; and they are constrained both semanti...

Introduction The availability of linguistically annotated corpora like the Penn Treebank has long proven beneficial for computational linguistics research. However, attempts have begun only recently to provide corpora with discourse information (c.f. e.g. URML (Reitter and Stede, 2003) or the Penn Discourse Treebank (PDTB) project at the University...

This paper describes a distributional approach to the semantics of verb-particle constructions (e.g. put up, make off ). We report first on a framework for implementing and evaluating such models. We then go on to report on the implementation of some techniques for using statistical models acquired from corpus data to infer the meaning of verb-part...

In this article we investigate logical metonymy, that is, constructions in which the argument of a word in syntax appears to be different from that argument in logical form (e.g., enjoy the book means enjoy reading the book, and easy problem means a problem that is easy to solve). The systematic variation in the interpretation of such constructions...

We present an overview of a comprehensive formal theory of the interpretation of sentential fragments, which has as components an empirically validated taxonomy, an analysis of the syntax and compositional semantics of fragments, and a formalisation of their contextual interpretation. We also briefly describe an implementation of this theory, and q...

Research on the discovery of terms from corpora has focused on word sequences whose recurrent occurrence in a corpus is indicative of their terminological sta- tus, and has not addressed the issue of discovering terms when data is sparse.

In this paper, we oer a semantic analysis of imperatives. We explore the eects of context on their interpretation, particularly on the content of the action to be performed, and whether or not the imperative is commanded. We demonstrate that by utilising a dynamic, discourse semantics which features rhetorical relations such as Narration, Elaborati...

Introduction Given the cansal and temporal relations between events in a knowledge base, what are the ways they can be described in text? Elsewhere, we have argued that during interpretation, the reader-hearer H must infer certain tempc ral information from knowledge about the world, language use and pragmatics. It is getterally agreed that process...

This paper motivates and describes a dynamic semantic theory of discourse interpretation called sdrt, which uses rhetorical relations to model the semantics/pragmatics interface. We describe the syntax and dynamic semantics of the language in which logical forms are represented, a separate but related language in which semantic underspeci - cation...

We describe the use of XML tokenisation, tagging and mark-up tools to prepare a corpus for parsing. Our techniques are generally applicable but here we focus on parsing Medline abstracts with the ANLT wide-coverage grammar.

This paper describes an extension of RUDI, a dialogue system component for `Resolving Underspecification with Discourse Information' (Schlangen et al., 2001). The extension handles the resolution of the intended meaning of non-sentential utterances that denote propositions or questions. Some researchers have observed that there are complex syntacti...

this paper allows any defaults to be overridden by defaults which are associated with more specific types: thus priority ordering reflects the type hierarchy ordering. (In 6.2, we will mention other possibilities for imposing a priority order on defaults.) Barring criterion 6, all of the above properties are necessary for making default unification...

This paper reports on a number of experiments which are designed to investigate the extent to which current nlp resources are able to syntactically and semantically analyse biomedical text. We address two tasks: parsing a real corpus with a hand-built wide-coverage grammar, producing both syntactic analyses and logical forms and automatically compu...

We describe the use of a suite of highly flexible XML-based NLP tools in a project for processing and interpreting text in the medical domain. The main aim of the paper is to demonstrate the central role that XML mark-up and XML NLP tools have played in the analysis process and to describe the resultant annotated corpus of MEDLINE abstracts. In add...

this paper allows any defaults to be overridden by defaults which are associated with more specific types: thus priority ordering reflects the type hierarchy ordering. (In Section 6.2, we will mention other possibilities for imposing a priority order on defaults.) Barring criterion 6, all of the above properties are necessary for making default uni...

This document describes cetp, an automated theorem prover for a quantifier-free, nonnegative fragment of common-sense entailment (Asher & Morreau 1991, Asher 1995), which is implemented in prolog

We investigate various contextual effects on text interpretation, and account for them by providing contextual constraints in a logical theory of text interpretation. On the basis of the way these constraints interact with the other knowledge sources, we draw some general conclusions about the role of domain-specific information, top-down and botto...

This paper is about the flow of inference between communicative intentions, discourse structure and the do- main during discourse processing. We augment a theory of discourse interpretation with a theory of distinct mental attitudes and reasoning about them, in order to provide an account of how the attitudes interact with reasoning about discourse...

We describe a formal framework for inter- pretation of words and compounds in a discourse context which integrates a symbolic lexicon/grammar. word-sense probabilities, and a pragmatic component. The approach is motivated by the need to handle productive word use. In this paper, we concentrate on compound nominals. We discuss the inadequacies of ap...

We offer a semantics and pragmatics of the pluperfect in narrative discourse. We ex- amine in a formal model of implicature, how the reader's knowledge about the discourse, Gricean-maxims and causation con- tribute to the meaning of the pluperfect.

This paper presents a formal account of the temporal interpretation of text. The distinct nat- ural interpretations of texts with similar syntax are explained in terms of defeasible rules charac- tcrising causal laws and Gricean-style pragmatic maxims. Intuitively compelling patterns of defea. sible entaJlment that are supported by the logic in whi...

We examine the role of temporal connec- tives in multi-sentence discourse. In cer- tain contexts, sentences containing temporal connectives that are equivalent in tern~ poral structure can fail to be equivalent in terms of discourse coherence. We account for this by offering a novel, formal mech~ anism for accommodating the presuppositions in tempo...

This paper describes aUDt ("Resolving Underspecification with Discourse Information"), a dialogue system component which computes automatically some aspects of the content of scheduling dialogues, particularly the intended denoration of the temporal expressions, the speech acts performed and the underlying goals. aUDt has a number of nice features:...

In this paper, we address several puzzles concerning speech acts,particularly indirect speech acts. We show how a formal semantictheory of discourse interpretation can be used to define speech actsand to avoid murky issues concerning the metaphysics of action. Weprovide a formally precise definition of indirect speech acts, includingthe subclass of...

This paper describes aUDt ("Resolving Underspecification with Discourse Information"), a dialogue system component which computes automatically some aspects of the content of scheduling dialogues, particularly the intended denoration of the temporal expressions, the speech acts performed and the underlying goals. aUDt has a number of nice features:...

We develop a framework for formalizing semantic construction within grammars expressed in typed feature structure logics, including HPSG. The approach provides an alternative to the lambda calculus; it maintains much of the desirable flexibility of unificationbased approaches to composition, while constraining the allowable operations in order to c...

We describe the use of XML tokenisation, tagging and mark-up tools to prepare a corpus for parsing. Our techniques are generally applicable but here we focus on parsing Medline abstracts with the ANLT wide-coverage grammar. Hand-crafted grammars inevitably lack coverage but many coverage failures are due to inadequacies of their lexicons. We descri...

We investigate various contextual eects on text interpretation, and account for them by providing contextual constraints in a logical theory of text interpretation. On the basis of the way these constraints interact with the other knowledge sources, we draw some general conclusions about the role of domain-specic information, top-down and bottom-up...

this paper, we examine anaphora in dialogue. Anaphora resolution is extremely complex, so for the sake of clarity we will focus on just two puzzles, illustrating them with a `minimal pair' of examples, in order to bring out the major features of the framework. The rst puzzle is that constraints on anaphora are not only sensitive to contextual infor...

Language is always generated and interpreted in a certain context, and the semantic, syntactic, and lexical properties of linguistic expressions reflect this. Interactive language understanding systems, such as language-based dialogue systems, therefore have to apply contextual information to interpret their inputs and to generate appropriate outpu...

In this paper we show that the dynamic interpretation techniques of Janssen (assignment modalities), Groenendijk and Stokhof (dynamic binding), and Hendriks (exibly scoping rules) enable a rigorous formulation of the semantics of intersentential anaphoric relationships, as well as of telescoping and periscoping phenomena in natural language. Keywor...

Default unification has been used in several linguistic applications. Most of them have utilised defaults at a meta-level, as part of an extended description language. We propose that allowing default unification to be a fully integrated part of a typed feature structure system requires default unification to be a binary, order independent function...

ormal semantics constitutes the framework of this thesis, and the aim is to characterise the semantics ( of the progressive, as it appears in sentence (1). 1) Max was running towards the station Among the problems is one known as the "imperfective paradox". According to intuitions, sentence ( (1) entails (2), but no entailment holds between (3) and...

In this paper, we offer a novel analysis of bridging, paying particular attention to definite descriptions. We argue that extant theories don't do justice to the way different knowledge resources interact. In line with Hobbs (1979), we claim that the rhetorical connections between the propositions introduced in the text plays an important part. But...

This paper focuses on metaphor and the interpretation of metaphor in a discourse setting. We propose constraints on their interpretation in terms of linguistic structures. Specifically, the constraints are based on a particular conception of the lexicon, where lexical entries have rich internal structure, and derivational processes or productivity...

Introduction Given the causal and temporal relations between events in a knowledge base, what are the ways they can be described in text? Elsewhere, we have argued that during interpretation, the reader-hearer H must infer certain temporal information from knowledge about the world, language use and pragmatics. It is generally agreed that processes...

We focus on the following question: given the causal and temporal relations between events in a knowledge base, what are the ways they can be described in extended text? We argue that we want to be able to generate laconic text, where certain temporal information remains implicit but pragmatically inferrable. An algorithm for generating laconic tex...

this paper, we'll consider three examples where this occurs: logical metonymy (e.g., enjoy the book means enjoy reading the book), adjectives (e.g., the interpretation of fast in fast car, fast motorway, fast typist etc.), and noun-verb agreement. We'll argue for a new version of default inheritance, which allows default results of lexical generali...

In this paper, we explore the interaction between lexical semantics and pragmatics. We argue that linguistic processing is informationally encapsulated and utilises relatively simple `taxonomic' lexical semantic knowledge. On this basis, defeasible lexical generalisations deliver defeasible parts of logical form. In contrast, pragmatic inference is...

In this paper, we offer a novel method for processing given information in discourse, paying particular attention to definite descriptions. We argue that extant theories don't do justice to the complexity of interaction between the knowledge resources that are used. In line with Hobbs (1979), we claim that discourse structure---as defined by the rh...

We discuss data involving the temporal structure of connected discourse. Questions are raised about the relation between clause order in discourse and causal order in the world, and about the coherence of certain discourses. We maintain that interpretation is contextually influenced by knowledge of the world and of pragmatics, and that the role of...

We examine the role of temporal connectives in multi-sentence discourse. In certain contexts, sentences containing temporal connectives that are equivalent in temporal structure can fail to be equivalent in terms of discourse coherence. We account for this by offering a novel, formal mechanism for accommodating the presuppositions in temporal subor...