Dialogue Management Research Articles

Conversation Graph: Data Augmentation, Training, and Evaluation for Non-Deterministic Dialogue Management

Task-oriented dialogue systems typically rely on large amounts of high-quality training data or require complex handcrafted rules. However, existing datasets are often limited in size con- sidering the complexity of the dialogues. Additionally, conventional training signal in- ference is not suitable for non-deterministic agent behavior, namely, considering multiple actions as valid in identical dialogue states. We propose the Conversation Graph (ConvGraph), a graph-based representation of dialogues that can be exploited for data augmentation, multi- reference training and evaluation of non- deterministic agents. ConvGraph generates novel dialogue paths to augment data volume and diversity. Intrinsic and extrinsic evaluation across three datasets shows that data augmentation and/or multi-reference training with ConvGraph can improve dialogue success rates by up to 6.4%.

An empirical assessment of deep learning approaches to task-oriented dialog management

Deep learning is providing very positive results in areas related to conversational interfaces, such as speech recognition, but its potential benefit for dialog management has still not been fully studied. In this paper, we perform an assessment of different configurations for deep-learned dialog management with three dialog corpora from different application domains and varying in size, dimensionality and possible system responses. Our results have allowed us to identify several aspects that can have an impact on accuracy, including the approaches used for feature extraction, input representation, context consideration and the hyper-parameters of the deep neural networks employed.

A novel focus encoding scheme for addressee detection in multiparty interaction using machine learning algorithms

Addressee detection is a fundamental task for seamless dialogue management and turn taking in human-agent interaction. Though addressee detection is implicit in dyadic interaction, it becomes a challenging task when more than two participants are involved. This article proposes multiple addressee detection models based on smart feature selection and focus encoding schemes. The models are trained using different machine learning and deep learning algorithms. This research work improves existing baseline accuracies for addressee prediction on two datasets. In addition, the article explores the impact of different focus encoding schemes in several addressee detection cases. Finally, an implementation strategy for addressee detection model in real-time is discussed.

Spoken dialogue management using neural networks: Basis of statistical dialogue management, and applying neural networks to the dialogue management

Spoken dialogue management using neural networks: Basis of statistical dialogue management, and applying neural networks to the dialogue management

A Language Tutoring Tool based on AI and Paraphrase Detection

A language tutoring tool (LTT) helps learning a language through casual human-like conversations. Natural language understanding (NLU) and natural language generation (NLG) are two key components of an LTT. In this paper, we propose a paraphrase detection algorithm that is used as the building block of the NLU. Our proposed tree-LSTM with a self-attention method for paraphrase detection shows accuracy of 87% with a lower parameter of 6.5m, which is much robust and lighter than the other existing paraphrase detection algorithms. Furthermore, we discuss an LTT prototype using the proposed algorithm with having some featured components like- message analysis, grammar detection, dialogue management, and response generation component. Each component is discussed in detail in the methodology section of this paper.

The AMIRO Social Robotics Framework: Deployment and Evaluation on the Pepper Robot.

Recent studies in social robotics show that it can provide economic efficiency and growth in domains such as retail, entertainment, and active and assisted living (AAL). Recent work also highlights that users have the expectation of affordable social robotics platforms, providing focused and specific assistance in a robust manner. In this paper, we present the AMIRO social robotics framework, designed in a modular and robust way for assistive care scenarios. The framework includes robotic services for navigation, person detection and recognition, multi-lingual natural language interaction and dialogue management, as well as activity recognition and general behavior composition. We present AMIRO platform independent implementation based on a Robot Operating System (ROS). We focus on quantitative evaluations of each functionality module, providing discussions on their performance in different settings and the possible improvements. We showcase the deployment of the AMIRO framework on a popular social robotics platform—the Pepper robot—and present the experience of developing a complex user interaction scenario, employing all available functionality modules within AMIRO.

Technical Aspects of Developing Chatbots for Medical Applications: Scoping Review.

BackgroundChatbots are applications that can conduct natural language conversations with users. In the medical field, chatbots have been developed and used to serve different purposes. They provide patients with timely information that can be critical in some scenarios, such as access to mental health resources. Since the development of the first chatbot, ELIZA, in the late 1960s, much effort has followed to produce chatbots for various health purposes developed in different ways.ObjectiveThis study aimed to explore the technical aspects and development methodologies associated with chatbots used in the medical field to explain the best methods of development and support chatbot development researchers on their future work.MethodsWe searched for relevant articles in 8 literature databases (IEEE, ACM, Springer, ScienceDirect, Embase, MEDLINE, PsycINFO, and Google Scholar). We also performed forward and backward reference checking of the selected articles. Study selection was performed by one reviewer, and 50% of the selected studies were randomly checked by a second reviewer. A narrative approach was used for result synthesis. Chatbots were classified based on the different technical aspects of their development. The main chatbot components were identified in addition to the different techniques for implementing each module.ResultsThe original search returned 2481 publications, of which we identified 45 studies that matched our inclusion and exclusion criteria. The most common language of communication between users and chatbots was English (n=23). We identified 4 main modules: text understanding module, dialog management module, database layer, and text generation module. The most common technique for developing text understanding and dialogue management is the pattern matching method (n=18 and n=25, respectively). The most common text generation is fixed output (n=36). Very few studies relied on generating original output. Most studies kept a medical knowledge base to be used by the chatbot for different purposes throughout the conversations. A few studies kept conversation scripts and collected user data and previous conversations.ConclusionsMany chatbots have been developed for medical use, at an increasing rate. There is a recent, apparent shift in adopting machine learning–based approaches for developing chatbot systems. Further research can be conducted to link clinical outcomes to different chatbot development techniques and technical characteristics.

Collaborative Dialogue Information Model

This paper concerns a task-oriented dialogue information management. Below we cover its main features – knowledge combining and intention detection. The approach is demonstrated in a step by step manner on one of sessions with a spoken timetable application.

Conversational ontology operator: patient-centric vaccine dialogue management engine for spoken conversational agents

BackgroundPreviously, we introduced our Patient Health Information Dialogue Ontology (PHIDO) that manages the dialogue and contextual information of the session between an agent and a health consumer. In this study, we take the next step and introduce the Conversational Ontology Operator (COO), the software engine harnessing PHIDO. We also developed a question-answering subsystem called Frankenstein Ontology Question-Answering for User-centric Systems (FOQUS) to support the dialogue interaction.MethodsWe tested both the dialogue engine and the question-answering system using application-based competency questions and questions furnished from our previous Wizard of OZ simulation trials.ResultsOur results revealed that the dialogue engine is able to perform the core tasks of communicating health information and conversational flow. Inter-rater agreement and accuracy scores among four reviewers indicated perceived, acceptable responses to the questions asked by participants from the simulation studies, yet the composition of the responses was deemed mediocre by our evaluators.ConclusionsOverall, we present some preliminary evidence of a functioning ontology-based system to manage dialogue and consumer questions. Future plans for this work will involve deploying this system in a speech-enabled agent to assess its usage with potential health consumer users.

Chatbot Kim: A Digital Tutor on AI. How Advanced Dialog Design Creates Better Conversational Learning Experiences

Conversational user interfaces, aka chatbots, offer new ways of interaction that can be used not only for task-led applications, but also for learning itself. Still, most conversational learning applications offer a predominantly one-directional dialog – either bot-led, with the user only confirming, or user-led, with the bot answering questions, but not going beyond. In contrast to these common approaches, learnbot Kim by time4you [1] not only conveys information, but wraps it in an equally entertaining and instructive chat, combining pre-defined dialog turns with a flexible dialog management. This article explains the design decisions made during the dialog development process and the underlying reasons. After having read this text, you will have a better idea of how dialog in conversational learning can be modeled to allow for a more natural conversational experience.

A proposal to strengthen public health surveillance systems on air pollution in Latin America and the Caribbean (LAC)

Air pollution is a major public health problem and a contributor to health inequities. It causes 300,000 attributable deaths in the Americas (41/100,000 in middle and low income countries, and 14/100,000 in high income countries). However, public health managers and health professionals do not have agency on emissions control or air quality monitoring. We will report the current status of air pollution surveillance systems in LAC, mainly based on the results of a self-assessment survey answered by technical professionals working in Ministries of health, Ministry of environment, or other relevant national institution or organization. We argue that a well-structured and reliable environmental public health system is essential to: measure the status and the progress of a country towards clean air for all based on a systematic collection, analysis and reporting of a selected set of health indicators; and provide the tools public health managers and other health professionals need to contribute in a multi-sectoral dialogue and influence air quality management and air quality policies that best could protect public health. Finally, we also comment on knowledge gaps and research needs.

Human-Robot Interaction Based on Dialog Management Using Sentence Similarity Comparison Method

Advances in developing dialogue systems regarding speech recognition, language understanding, and speech synthesis. Dialogue systems to support human interaction with a robot efficiently by using spoken language. Facilities that provide convenience in carrying out daily activities for someone, such as older people, are necessary. The existence of Human-Robot Interaction (HRI), so that this interaction can give orders to the robot to do work that cannot be done by humans. This study presents a dialogue management system for HRI with a comparison sentence similarity method between TF-IDF (Term Frequency-Inverse Document Frequency) Cosine Similarity Algorithm and Jaccard Coefficient and using Finite State Machine (FSM). Dialogue Management is a way to find the response of the answer. When the user says something or in other words, is responsible for managing the flow of the conversation to command the robot. TF-IDF is used to give the weight of the term relationship and comparison between Cosine Similarity and Jaccard Coefficient for comparison method to determine the classification of similarity sentences from the dialogue manager to improve the intent of the dialogue, for the FSM method to set the sequence flow dialogue. We use Google Cloud Speech API as an engine for speech to text using Kinect V2 as an audio sensor. There are eight scenarios created in this system. The speech recognition process using Google Speech for an average of 2.62 seconds shows a reasonably fast response. TF-IDF Cosine Similarity method can produce enough accuracy of 97.43%, and Jaccard Coefficient indicates an accuracy level of 91.57%. The state of the FSM method can be considered as an efficient structure for building dialogue management.

Adaptive dialogue management using intent clustering and fuzzy rules

AbstractConversational systems have become an element of everyday life for billions of users who use speech‐based interfaces to services, engage with personal digital assistants on smartphones, social media chatbots, or smart speakers. One of the most complex tasks in the development of these systems is to design the dialogue model, the logic that provided a user input selects the next answer. The dialogue model must also consider mechanisms to adapt the response of the system and the interaction style according to different groups and user profiles. Rule‐based systems are difficult to adapt to phenomena that were not taken into consideration at design‐time. However, many of the systems that are commercially available are based on rules, and so are the most widespread tools for the development of chatbots and speech interfaces. In this article, we present a proposal to: (a) automatically generate the dialogue rules from a dialogue corpus through the use of evolving algorithms, (b) adapt the rules according to the detected user intention. We have evaluated our proposal with several conversational systems of different application domains, from which our approach provided an efficient way for adapting a set of dialogue rules considering user utterance clusters.

Adaptable Conversational Machines

In recent years we have witnessed a surge in machine learning methods that provide machines with conversational abilities. Most notably, neural‐network‐based systems have set the state of the art for difficult tasks such as speech recognition, semantic understanding, dialogue management, language generation, and speech synthesis. Still, unlike for the ancient game of Go for instance, we are far from achieving human‐level performance in dialogue. The reasons for this are numerous. One property of human‐human dialogue that stands out is the infinite number of possibilities of expressing oneself during the conversation, even when the topic of the conversation is restricted. A typical solution to this problem was scaling‐up the data. The most prominent mantra in speech and language technology has been “There is no data like more data.” However, the researchers now are focused on building smarter algorithms — algorithms that can learn efficiently from just a few examples. This is an intrinsic property of human behavior: an average human sees during their lifetime a fraction of data that we nowadays present to machines. A human can even have an intuition about a solution before ever experiencing an example solution. The human‐inspired ability to adapt may just be one of the keys in pushing dialogue systems toward human performance. This article reviews advancements in dialogue systems research with a focus on the adaptation methods for dialogue modeling, and ventures to have a glance at the future of research on adaptable conversational machines.

Adopter quand on a une descendance biologique

Why adopt when you have biological descendants, and why adopt ostensibly by choosing a child from a black African country? This is the question we address in this article from a doctoral thesis in psychology supported in 2017 at the University of Lorraine. Although adoption has been the subject of much research, the scientific literature on “international adoption”, which will occupy us here, and which, at least in France, has tended to dry up in recent years, is very little provided. That is why, in order to open up avenues of research, we wanted to engage in an exploratory study of the appropriation that couples give in the after-the-fact to this adventure in which they engaged. Five couples were therefore interviewed in non-directive research interviews (Blanchet, 2003) conducted by the clinical psychologist involved in this thesis. This style of adoption is basically a “couple affair”, each couple being heard in three separate meetings in about a week. The first interview took place with the couple, the next two involving just the wife and then the husband. The interviews were recorded, then transcribed verbatim, finally analyzed thematically, through a relatively informal content analysis, and then pragmatically, using tools for analysis of the dialogue relating to enunciation (for example on people's deixis [Benveniste, 1966]) and dialogue management (e.g. the management of alternation and dialogical sequences (Trognon and Brassac, 1993; Trognon and Coulon, 2001; Trognon and Batt, 2010). This approach has identified many traits specific to those involved in the adventure of “international adoption”. Related, for example, to the individual and interindividual history of the members of the couple, the place that is devolved to the adopted child in the couple's imagination, and the commitment that both parents make in the common project. Their knowledge should prove to be useful to any professional who meets families who have experienced an international adoption.

A framework for building closed-domain chat dialogue systems

This paper presents HRIChat, a framework for developing closed-domain chat dialogue systems. Being able to engage in chat dialogues has been found effective for improving communication between humans and dialogue systems. This paper focuses on closed-domain systems because they would be useful when combined with task-oriented dialogue systems in the same domain. HRIChat enables domain-dependent language understanding so that it can deal well with domain-specific utterances. In addition, HRIChat makes it possible to integrate state transition network-based dialogue management and reaction-based dialogue management. FoodChatbot, which is an application in the food and restaurant domain, has been developed and evaluated through a user study. Its results suggest that reasonably good systems can be developed with HRIChat. This paper also reports lessons learned from the development and evaluation of FoodChatbot.

Has multiculturalism failed? Let's start the era of interculturalism for facing diversity issues

ABSTRACT In view of the primary challenges facing contemporary society, such as globalisation and interdependence, the author reflects on the question of what the most appropriate educational policy and pedagogical approach is: transcultural education, multicultural education, or intercultural education. Based on historical developments in the US and in Europe, he first presents a terminological and semantic clarification. Then, considering the existing terminological confusion and misunderstandings across the world and disciplines, he underlines the urgency to develop a shared terminology. He proposes to use the adjective ‘multicultural’ in descriptive terms (societies are multicultural) and applies multicultural education as a strategy for improving knowledge about and respect for ethnic and cultural diversities. The concept ‘intercultural’ should be utilised in reference to intervention, where intercultural education is viewed as an activity in which one engages in interaction: encounter, contact, dialogue, and conflict management. In his opinion, nowadays, intercultural education also has to include new challenges and different forms of diversity, including neoliberalism, climate change, power, and special needs.

A hierarchical approach for efficient multi-intent dialogue policy learning

This paper proposes a hierarchical method for learning an efficient Dialogue Management (DM) strategy for task-oriented conversations serving multiple intents of a domain. Deep Reinforcement Learning (DRL) networks specializing in individual intents communicate with each other, having the capability of sharing overlapping information across intents. The sharing of information across state space and the presence of global slot tracker prohibits the agent to reask known information. Thus, the system is able to handle sub-dialogues based on subset of intents covered by different Reinforcement Learning (RL) models, thereby, completing the dialogue without again asking already provided information common across intents. The developed system has been demonstrated for “Air Travel” domain. The experimental results indicate that the developed system is efficient, scalable and can serve multiple intents based dialogues adequately. The proposed system when applied to 5-intent dialogue systems attains an improvement of 41% in terms of dialogue length as compared to a single-intent based system serving the same 5-intents.

A comparative review on deep learning models for text classification

<p>Text classification is a fundamental task in several areas of natural language processing (NLP), including words semantic classification, sentiment analysis, question answering, or dialog management. This paper investigates three basic architectures of deep learning models for the tasks of text classification: Deep Belief Neural (DBN), Convolutional Neural Network (CNN) and Recurrent Neural Network (RNN), these three main types of deep learning architectures, are largely explored to handled various classification tasks. DBN have excellent learning capabilities to extracts highly distinguishable features and good for general purpose. CNN have supposed to be better at extracting the position of various related features while RNN is modeling in sequential of long-term dependencies. This paper work shows the systematic comparison of DBN, CNN, and RNN on text classification tasks. Finally, we show the results of deep models by research experiment. The aim of this paper to provides basic guidance about the deep learning models that which models are best for the task of text classification.</p>

Toward Personalized Scaffolding and Fading of Motivational Support in L2 Learner–Dialogue Agent Interactions: An Exploratory Study

This article proposes a computer-based approach to effectively enhance second language learners' willingness to communicate in the target language. To do so, we implemented a conversational agent embedding a dialogue management model based on two conversational strategies (i.e., communication strategies and affective backchannels), serving as scaffolds for enhancing learners' willingness to communicate in the target language. Here, we report on differences observed among second language learners' preferences for both conversational strategies according to their initial level of willingness to communicate and on variations of their willingness with respect to such differences. Although we found that most students generally preferred a combination of both strategies, learners' preferences and the effects of the support provided by these strategies varied according to their level of willingness to communicate. Learners with lower willingness to communicate tended to prefer affective backchannels, whereas those with higher willingness to communicate seemed to favor communication strategies. These results were consistent with post-test results, which showed that learners' expected willingness to communicate tended to be higher after interacting with systems embedding their preferred strategies. In sum, these results are preliminary evidence of the meaningfulness of accounting for such learners' preferences in adaptively using and fading the strategies employed by conversational agents to motivate second language learners to communicate in the target language.