Systems Engineering and Conversational Agents

Abstract

Abstract. This chapter describes Conversational Agents (CAs) in the context of Systems Engineering. A CA is a computer program which interacts with a user through natural language dialogue and provides some form of service. CA technology has two points of interest to systems engineers: the use of systems engineering techniques in CA research and the application of CAs in projects development. CAs offer the opportunity to automate more complex applications than are feasible with conventional web interfaces. Currently such applications require a human expert in the domain to mediate between the user and the application. The CA effectively replaces the human expert. This chapter reviews the current capabilities of various CA technologies, outlines a development methodology for systems engineering practitioners interested in developing real world applications and suggests a number of directions for systems engineers who wish to participate in CA research.Keywords: Conversational agent, systems engineering, dialogue, evaluation, methodology, semantic similarity, short text Section 1 IntroductionIn the early decades of computing ordinary people did not have any interaction with computers at all, any information that was needed for an application was entered by specialised clerks. This was followed by a period (largely driven by the internet) in which users could drive simple applications by entering simple facts and figures directly. Recently an explosion of internet use has demanded that ordinary people interact with complex applications, and the trend is for these to be of increasing complexity. Such applications would normally require a domain expert to interview the user to obtain the necessary information. Automating such processes represents a serious challenge to computing practitioners. This chapter describes a possible way forward, Conversational Agents (CAs). CA technology has two points of interest to systems engineers: the use of systems engineering techniques in CA research and the application of CAs in projects development. A CA is a computer program which interacts with a user through natural language dialogue and provides some form of service.

Goal Orientated Conversational Agents: Applications to Benefit Society

“…This section describes an architecture developed by the Intelligent Systems Group at MMU for constructing GO-CAs [2]. The methodology is designed so the user can approach the GO-CA with a problem or a request for information which can then lead to interactive dialogue between the user and the GO-CA until the user's goal is achieved.…”

“…A further strength of a GO-CA is that it will engage in extended dialogue, during the course of which it will appear to have mental states that are directed at states of affairs in the world e.g. beliefs, desires, hopes, perception, intention etc [2].The GO-CA is a mixed-initiative system (from time to time either the human or the agent may take control of the conversation) [2]. However, due to the nature of the GO-CA, if the human diverts the conversation from the goal, the GO-CA will always attempt to get the conversational back on track using the rule base and knowledge obtained from attribute values captured during the conversation.…”

“…Figure 1 shows the generic modular architecture for a typical GO-CA [3]. This is intended to take on challenging real-world applications in which the user may present adversarial, disruptive or deceptive behavior at times during the conversation [2]. Each component within the architecture will now be briefly described:  The discourse manager is the central component of the GO-CA and provides communication with the problem specific expert system (rule base), dynamic graphical user interface (GUI) and the Dialogue Agent (DA).…”

“…A Goal-Oriented CA (GO-CA) is a type of conversational agent which has a deep strategic purpose which enables it to direct a conversation to achieve a goal [2]. Thus the GO-CA [2,3] may spend more time leading the conversation and asking questions than the human participant which makes it different from systems such as chatterbots and traditional conversational agents.…”

“…Thus the GO-CA [2,3] may spend more time leading the conversation and asking questions than the human participant which makes it different from systems such as chatterbots and traditional conversational agents. Chatterbots [4] attempt to prolong social conversational with a human for as long as possible generally using pointless chat [2]. Traditional conversational agents either focus on the embodied component (e.g.…”