We address the problem of dynamic user modelling for referring expression generation in spoken dialogue systems, i.e how a spoken dialogue system should choose referring expressions to refer to domain entities to users with different levels of domain expertise, whose domain knowledge is initially unknown to the system. We approach this problem using a statistical planning framework: Reinforcement Learning techniques in Markov Decision Processes (MDP). We present a new reinforcement learning framework to learn user modelling strategies for adaptive referring expression generation (REG) in resource scarce domains (i.e. where no large corpus exists for learning). As a part of the framework, we present novel user simulation models that are sensitive to the referring expressions used by the system and are able to simulate users with different levels of domain knowledge. Such models are shown to simulate real user behaviour more closely than baseline user simulation models. In contrast to previous approaches to user adaptive systems, we do not assume that the user’s domain knowledge is available to the system before the conversation starts. We show that using a small corpus of non-adaptive dialogues it is possible to learn an adaptive user modelling policy in resource scarce domains using our framework. We also show that the learned user modelling strategies performed better in terms of adaptation than hand-coded baselines policies on both simulated and real users. With real users, the learned policy produced around 20% increase in adaptation in comparison to the best performing hand-coded adaptive baseline. We also show that adaptation to user’s domain knowledge results in improving task success (99.47% for learned policy vs 84.7% for hand-coded baseline) and reducing dialogue time of the conversation (11% relative difference). This is because users found it easier to identify domain objects when the system used adaptive referring expressions during the conversations.

A Spoken Dialogue System's (SDS's) dialogue strategy specifies which action it will take depending on its representation of the current dialogue context. Designing it by hand involves anticipating how users will interact with the system, and/or repeated testing and refining, and so can be a difficult, time-consuming task. Since SDSs inevitably make understanding errors, a particularly important issue is how to design ``repair strategies'', the parts of the dialogue strategy which attempt to get the dialogue ``back-on-track'' following these errors. To try to produce better dialogue strategies with less time and effort, previous researchers have modelled a dialogue strategy as a sequential decision problem called a Markov Decision Process (MDP), and then applied Reinforcement Learning (RL) algorithms to example training dialogues to generate dialogue strategies automatically. More recent research has used training dialogues conducted with simulated rather than real users and learned which action to take in all dialogue contexts, (a ``full'' as opposed to a ``partial'' dialogue strategy) - simulated users allow more training dialogues to be generated, and the exploration of new dialogue contexts not present in an original dataset. As yet however, limited insight has been provided as to which dialogue contextual features are important to include in the MDP and why. Indeed, a full dialogue strategy has not been learned from training dialogues with a realistic probabilistic user simulation derived from real user data, and then shown to work well with real users. This thesis investigates the value of adding new linguistically-motivated contextual features to the MDP when using RL to learn full dialogue strategies for SDSs. These new features are recent Dialogue Acts (DAs). DAs indicate the role or intention of an utterance in a dialogue e.g. ``provide-information'', an utterance being a complete unit of a speaker's speech, often bounded by silence. An accurate probabilistic user simulation learned from real user data is used for generating training dialogues, and the recent DAs are shown to improve performance in testing in simulation and with real users. With real users, performance is also better than other competing learned and hand-crafted strategies. Analysis of the strategies, and further simulation experiments show how the DAs improve performance through better repair strategies. The main findings are expected to apply to SDSs in general - indeed our strategies are learned and tested on real users in different domains, (flight-booking versus tourist information). Comparisons are also made to recent research which focuses on handling understanding errors in SDSs, but which does not use RL or user simulations.

This thesis presents a development of an English spoken dialogue system based on the Alex dialogue system framework. The work describes a component adaptation of the framework for a different domain and language. The system provides public transport information in New York. This work involves creating a statistical model and the deployment of custom Kaldi speech recognizer. Its performance was better in comparison with the Google Speech API. The comparison was based on a subjective user satisfaction acquired by crowdsourcing. Powered by TCPDF (www.tcpdf.org)

From a system developer's perspective, designing a spoken dialogue system can be a time-consuming and difficult process. A developer may spend a lot of time anticipating how a potential user might interact with the system and then deciding on the most appropriate system response. These decisions are encoded in a dialogue strategy, essentially a mapping between anticipated user inputs and appropriate system outputs. To reduce the time and effort associated with developing a dialogue strategy, recent work has concentrated on modelling the development of a dialogue strategy as a sequential decision problem. Using this model, reinforcement learning algorithms have been employed to generate dialogue strategies automatically. These algorithms learn strategies by interacting with simulated users. Some progress has been made with this method but a number of important challenges remain. For instance, relatively little success has been achieved with the large state representations that are typical of real-life systems. Another crucial issue is the time and effort associated with the creation of simulated users. In this thesis, I propose an alternative to existing reinforcement learning methods of dialogue strategy development. More specifically, I explore how XCS, an evolutionary reinforcement learning algorithm, can be used to find dialogue strategies that cover large state spaces. Furthermore, I suggest that hand-coded simulated users are sufficient for the learning of useful dialogue strategies. I argue that the use of evolutionary reinforcement learning and hand-coded simulated users is an effective approach to the rapid development of spoken dialogue strategies. Finally, I substantiate this claim by evaluating a learned strategy with real users. Both the learned strategy and a state-of-the-art hand-coded strategy were integrated into an end-to-end spoken dialogue system. The dialogue system allowed real users to make flight enquiries using a live database for an Edinburgh-based airline. The performance of the learned and hand-coded strategies were compared. The evaluation results show that the learned strategy performs as well as the hand-coded one (81% and 77% task completion respectively) but takes much less time to design (two days instead of two weeks). Moreover, the learned strategy compares favourably with previous user evaluations of learned strategies.

This thesis addresses the question of how speakers adapttheir language when they interact with a spoken dialoguesystem. In human–human dialogue, people continuously adaptto their conversational partners at different levels. Wheninteracting with computers, speakers also to some extent adapttheir language to meet (what they believe to be) theconstraints of the dialogue system. Furthermore, if a problemoccurs in the human–computer dialogue, patterns oflinguistic adaptation are often accentuated. In this thesis, we used an empirical approach in which aseries of corpora of human–computer interaction werecollected and analyzed. The systems used for data collectionincluded both fully functional stand-alone systems in publicsettings, and simulated systems in controlled laboratoryenvironments. All of the systems featured animated talkingagents, and encouraged users to interact using unrestrictedspontaneous language. Linguistic adaptation in the corpora wasexamined at the phonetic, prosodic, lexical, syntactic andpragmatic levels. Knowledge about users’linguistic adaptations can beuseful in the development of spoken dialogue systems. If we areable to adequately describe their patterns of occurrence (atthe different linguistic levels at which they occur), we willbe able to build more precise user models, thus improvingsystem performance. Our knowledge of linguistic adaptations canbe useful in at least two ways: first, it has been shown thatlinguistic adaptations can be used to identify (andsubsequently repair) errors in human–computer dialogue.Second, we can try to subtly influence users to behave in acertain way, for instance by implicitly encouraging a speakingstyle that improves speech recognition performance.

<p>This thesis addresses the question of how speakers adapttheir language when they interact with a spoken dialoguesystem. In human–human dialogue, people continuously adaptto their conversational partners at different levels. Wheninteracting with computers, speakers also to some extent adapttheir language to meet (what they believe to be) theconstraints of the dialogue system. Furthermore, if a problemoccurs in the human–computer dialogue, patterns oflinguistic adaptation are often accentuated.</p><p>In this thesis, we used an empirical approach in which aseries of corpora of human–computer interaction werecollected and analyzed. The systems used for data collectionincluded both fully functional stand-alone systems in publicsettings, and simulated systems in controlled laboratoryenvironments. All of the systems featured animated talkingagents, and encouraged users to interact using unrestrictedspontaneous language. Linguistic adaptation in the corpora wasexamined at the phonetic, prosodic, lexical, syntactic andpragmatic levels.</p><p>Knowledge about users’linguistic adaptations can beuseful in the development of spoken dialogue systems. If we areable to adequately describe their patterns of occurrence (atthe different linguistic levels at which they occur), we willbe able to build more precise user models, thus improvingsystem performance. Our knowledge of linguistic adaptations canbe useful in at least two ways: first, it has been shown thatlinguistic adaptations can be used to identify (andsubsequently repair) errors in human–computer dialogue.Second, we can try to subtly influence users to behave in acertain way, for instance by implicitly encouraging a speakingstyle that improves speech recognition performance.</p>

Recent advancements in the area of spoken language processing and the wide acceptance of portable devices, have attracted signicant interest in spoken dialogue systems.These conversational systems are man-machine interfaces which use natural language (speech) as the medium of interaction.In order to conduct dialogues, computers must have the ability to decide when and what information has to be exchanged with the users. The dialogue management module is responsible to make these decisions so that the intended task (such as ticket booking or appointment scheduling) can be achieved.Thus learning a good strategy for dialogue management is a critical task.In recent years reinforcement learning-based dialogue management optimization has evolved to be the state-of-the-art. A majority of the algorithms used for this purpose needs vast amounts of training data.However, data generation in the dialogue domain is an expensive and time consuming process. In order to cope with this and also to evaluatethe learnt dialogue strategies, user modelling in dialogue systems was introduced. These models simulate real users in order to generate synthetic data.Being computational models, they introduce some degree of modelling errors. In spite of this, system designers are forced to employ user models due to the data requirement of conventional reinforcement learning algorithms can learn optimal dialogue strategies from limited amount of training data when compared to the conventional algorithms. As a consequence of this, user models are no longer required for the purpose of optimization, yet they continue to provide a fast and easy means for quantifying the quality of dialogue strategies. Since existing methods for user modelling are relatively less realistic compared to real user behaviors, the focus is shifted towards user modelling by means of inverse reinforcement learning. Using experimental results, the proposed method's ability to learn a computational models with real user like qualities is showcased as part of this work.

Cette thèse s’inscrit dans le cadre de la recherche sur les systèmes de dialogue. Ce document propose d’apprendre le comportement d’un système à partir d’un ensemble de dialogues annotés. Le système apprend un comportement optimal via l’apprentissage par renforcement. Nous montrons qu’il n’est pas nécessaire de définir une représentation de l’espace d’état ni une fonction de récompense. En effet, ces deux paramètres peuvent être appris à partir du corpus de dialogues annotés. Nous montrons qu’il est possible pour un développeur de systèmes de dialogue d’optimiser la gestion du dialogue en définissant seulement la logique du dialogue ainsi qu’un critère à maximiser (par exemple, la satisfaction utilisateur). La première étape de la méthodologie que nous proposons consiste à prendre en compte un certain nombre de paramètres de dialogue afin de construire une représentation de l’espace d’état permettant d’optimiser le critère spécifié par le développeur. Par exemple, si le critère choisi est la satisfaction utilisateur, il est alors important d’inclure dans la représentation des paramètres tels que la durée du dialogue et le score de confiance de la reconnaissance vocale. L’espace d’état est modélisé par une mémoire sparse distribuée. Notre modèle, Genetic Sparse Distributed Memory for Reinforcement Learning (GSDMRL), permet de prendre en compte de nombreux paramètres de dialogue et de sélectionner ceux qui sont importants pour l’apprentissage par évolution génétique. L’espace d’état résultant ainsi que le comportement appris par le système sont aisément interprétables. Dans un second temps, les dialogues annotés servent à apprendre une fonction de récompense qui apprend au système à optimiser le critère donné par le développeur. A cet effet, nous proposons deux algorithmes, reward shaping et distance minimisation. Ces deux méthodes interprètent le critère à optimiser comme étant la récompense globale pour chaque dialogue. Nous comparons ces deux fonctions sur un ensemble de dialogues simulés et nous montrons que l’apprentissage est plus rapide avec ces fonctions qu’en utilisant directement le critère comme récompense finale. Nous avons développé un système de dialogue dédié à la prise de rendez-vous et nous avons collecté un corpus de dialogues annotés avec ce système. Ce corpus permet d’illustrer la capacité de mise à l’échelle de la représentation de l’espace d’état GSDMRL et constitue un bon exemple de système industriel sur lequel la méthodologie que nous proposons pourrait être appliquée / This document proposes to learn the behaviour of the dialogue manager of a spoken dialogue system from a set of rated dialogues. This learning is performed through reinforcement learning. Our method does not require the definition of a representation of the state space nor a reward function. These two high-level parameters are learnt from the corpus of rated dialogues. It is shown that the spoken dialogue designer can optimise dialogue management by simply defining the dialogue logic and a criterion to maximise (e.g user satisfaction). The methodology suggested in this thesis first considers the dialogue parameters that are necessary to compute a representation of the state space relevant for the criterion to be maximized. For instance, if the chosen criterion is user satisfaction then it is important to account for parameters such as dialogue duration and the average speech recognition confidence score. The state space is represented as a sparse distributed memory. The Genetic Sparse Distributed Memory for Reinforcement Learning (GSDMRL) accommodates many dialogue parameters and selects the parameters which are the most important for learning through genetic evolution. The resulting state space and the policy learnt on it are easily interpretable by the system designer. Secondly, the rated dialogues are used to learn a reward function which teaches the system to optimise the criterion. Two algorithms, reward shaping and distance minimisation are proposed to learn the reward function. These two algorithms consider the criterion to be the return for the entire dialogue. These functions are discussed and compared on simulated dialogues and it is shown that the resulting functions enable faster learning than using the criterion directly as the final reward. A spoken dialogue system for appointment scheduling was designed during this thesis, based on previous systems, and a corpus of rated dialogues with this system were collected. This corpus illustrates the scaling capability of the state space representation and is a good example of an industrial spoken dialogue system upon which the methodology could be applied

Title: Novel Methods for Natural Language Generation in Spoken Dialogue Systems Author: Ondřej Dušek Department: Institute of Formal and Applied Linguistics Supervisor: Ing. Mgr. Filip Jurčíček, Ph.D., Institute of Formal and Applied Linguistics Abstract: This thesis explores novel approaches to natural language generation (NLG) in spoken dialogue systems (i.e., generating system responses to be presented the user), aiming at simplifying adaptivity of NLG in three respects: domain portability, language portability, and user-adaptive outputs. Our generators improve over state-of-the-art in all of them: First, our gen- erators, which are based on statistical methods (A* search with perceptron ranking and sequence-to-sequence recurrent neural network architectures), can be trained on data without fine-grained semantic alignments, thus simplifying the process of retraining the generator for a new domain in comparison to previous approaches. Second, we enhance the neural-network-based gener- ator so that it takes preceding dialogue context into account (i.e., user's way of speaking), thus producing user-adaptive outputs. Third, we evaluate sev- eral extensions to the neural-network-based generator designed for producing output in morphologically rich languages, showing improvements in Czech generation. In...

This thesis focuses on the problem of scalable optimization of dialogue behaviour in speech-based conversational systems using reinforcement learning. Most previous investigations in dialogue strategy learning have proposed flat reinforcement learning methods, which are more suitable for small-scale spoken dialogue systems. This research formulates the problem in terms of Semi-Markov Decision Processes (SMDPs), and proposes two hierarchical reinforcement learning methods to optimize sub-dialogues rather than full dialogues. The first method uses a hierarchy of SMDPs, where every SMDP ignores irrelevant state variables and actions in order to optimize a sub-dialogue. The second method extends the first one by constraining every SMDP in the hierarchy with prior expert knowledge. The latter method proposes a learning algorithm called 'HAM+HSMQ-Learning', which combines two existing algorithms in the literature of hierarchical reinforcement learning. Whilst the first method generates fully-learnt behaviour, the second one generates semi-learnt behaviour. In addition, this research proposes a heuristic dialogue simulation environment for automatic dialogue strategy learning. Experiments were performed on simulated and real environments based on a travel planning spoken dialogue system. Experimental results provided evidence to support the following claims: First, both methods scale well at the cost of near-optimal solutions, resulting in slightly longer dialogues than the optimal solutions. Second, dialogue strategies learnt with coherent user behaviour and conservative recognition error rates can outperform a reasonable hand-coded strategy. Third, semi-learnt dialogue behaviours are a better alternative (because of their higher overall performance) than hand-coded or fully-learnt dialogue behaviours. Last, hierarchical reinforcement learning dialogue agents are feasible and promising for the (semi) automatic design of adaptive behaviours in larger-scale spoken dialogue systems. This research makes the following contributions to spoken dialogue systems which learn their dialogue behaviour. First, the Semi-Markov Decision Process (SMDP) model was proposed to learn spoken dialogue strategies in a scalable way. Second, the concept of 'partially specified dialogue strategies' was proposed for integrating simultaneously hand-coded and learnt spoken dialogue behaviours into a single learning framework. Third, an evaluation with real users of hierarchical reinforcement learning dialogue agents was essential to validate their effectiveness in a realistic environment.