Approaches for Dialog Management in Conversational Agents

Abstract

Dialog agents, like digital assistants and automated chat interfaces (e.g., chatbots), are becoming more and more popular as users adapt to conversing with their devices as they do with humans. In this paper, we present approaches and available tools for dialog management (DM), a component of dialog agents that handles dialog context and decides the next action for the agent to take. In this paper, we establish an overview of the field of DM, compare approaches and state-of-the-art tools in industry and research work on a set of dimensions, and identify directions for further research work. & THE DREAM OF a human-like highly intelligent computer assistant has been presented in many science fiction movies like Hal in "2001: A Space Odyssey" (1968), Samantha in "Her" (2013), and Jarvis in "Iron Man" (2013). Recent advances in automatic speech recognition systems, machine learning, and artificial intelligence enabled the advent of personal assistants like Google Assistant, Siri, and Alexa, first on smartphones and lately on home speakers and other devices. These might make the reality seem very close to the fiction in the movies. However, while the assistants are capable of executing small tasks, the richness and quality of their dialogs are not comparable to the ones of humans: interactions are still simple, short, and constrained by a limited vocabulary, thus forcing users to adjust to the system's capabilities. Digital assistants are part of a larger group of dialog agents which include: voice user interfaces (or spoken dialog systems), text-based agents and embodied conversational agents, 1 [Chapter 4]. Historically, dialog agents aimed to simulate human conversation. The first examples of textbased agents are ELIZA 2 which acted as a Rogerian psychotherapist, and PARRY 3 simulating a paranoid schizophrenic. This was possible with extensive rule-sets and structured question-answer sets. With advances in natural language processing

Robotic Telemedicine for Mental Health: A Multimodal Approach to Improve Human-Robot Engagement

“…To create natural, believable interactions between intelligent virtual assistants and humans, understanding the context of conversations is of utmost importance ( Harms et al, 2018 ). Therefore, for each dialogue skill designed several context variables were programmed (i.e., information that is stored during the dialogue), such as the user’s name, mood, time of day, or location.…”

“…This technology started in the 1960s with ELIZA ( Weizenbaum, 1966 ), which held text-based conversations with users acting as a psychotherapist. A recent body of work ( Romero et al, 2017 ; Harms et al, 2018 ) has provided novel approaches for the development of conversational agents for increased user engagement. Along similar lines, there has been work on the development of embodied conversational agents—virtual animated characters, usually with the appearance of a human-like avatar, capable of understanding multimodal utterances, such as voice, gestures, and emotion ( Griol et al, 2019 ).…”

Chatbot Integration in Few Patterns

“…To disambiguate similar intents or infer values of slots, additional information, the dialog context, may be used (e.g., if the chatbot already knows the location of the user, it does not need to ask for it in order to provide a localized weather forecast). The dialog control is designed either explicitly by defining conversation flows or derived from previous conversations, or using a combination of both techniques (refer to Harms et al 2 and Hussain et al 3 for more details on chatbots design and architecture). We illustrate these concepts in Figure 1a.…”

“…So far, chatbot development has been studied considering different aspects of design, such as interaction model, application domain, goal-orientation, and dialog management 3 . Existing surveys have analyzed popular chatbot systems and chatbot frameworks (e.g., Harms et al 2 ). All these classifications follow a white-box approach and focus on the ingredients that define the internals of a chatbot, which translate into conversational capabilities and, eventually, user experience.…”

Don't Believe The Hype!

“…Major industry platforms (e.g., Google's Dialogflow [7]; Amazon's Lex [12]; IBM's Watson Assistant [2]) take a straightforward frame-based approach to DM, managing context to ensure that all entities or 'slots' are filled prior to fulfilling an intent. (Harms et al provide a detailed comparison of DM approaches in both commercial and research tools [11]).…”