Abstract
Chatbots are rapidly growing and becoming prevalent in many applications, including healthcare, education, and consumer services. As such, it is crucial to design chatbots with usability in mind. Traditional usability heuristics call for error prevention, consistency, and task efficiency. The heuristics are suited for all types of user interfaces, including chatbots. However, chatbots have their unique challenges, and thus, recently, some researchers have developed sets of usability guidelines specifically for chatbots. However, there is a shortage of studies that summarize the latest advances in chatbot usability design and assessment. As such, this chapter covers the existing general and chatbot-specific usability heuristics and examples of usage. Moreover, the chapter presents the recent developments in chatbot design techniques and challenges.
TopIntroduction
Chatbots, also known as conversational agents, facilitate the interaction of humans with computers using natural language by applying natural language processing (NLP) (Bradeško & Mladenić, 2012). Chatbots are rapidly becoming prevalent as they imitate human conversations and thus automate service, and they are now utilized in various areas, including healthcare (Oh et al., 2017), consumer services (Xu et al., 2017), education (Kuhail et al., 2022), and academic advising (Kuhail et al., 2022).
Despite the popularity of chatbots in various fields and a plethora of studies on chatbots in academia, there is a lack of studies summarizing the recent development of chatbots' usability and user experience in terms of design and usability evaluation. In essence, usability assesses how easy a user interface is to use (Nathoo et al., 2019). Traditionally, Schneiderman et al. (2016), Sugisaki and Bleiker (2020), and Nielsen and Molich (1990) identified generic usability heuristics as useful for guiding and evaluating user interfaces. However, since they are generic, the usability heuristics are susceptible to different interpretations by designers (Sugisaki & Bleiker, 2020), especially for conversational interfaces, which have unique characteristics such as sequential communication and freedom of interaction and initiative (Shneiderman et al., 2016). Consequently, various researchers contributed usability heuristics to guide the design and assessment of conversational user interfaces (Bos et al., 1999) (Murad et al., 2019) (Sugisaki & Bleiker, 2020). These heuristics are established on the generic usability heuristics together with particularity connected to conversation and language studies. Other techniques and strategies have been used for evaluating chatbots, including log analysis and observations (Kawasaki et al., 2020) and surveys (Xiao et al., 2020).
Key Terms in this Chapter
Lexical Structure: The syntax or grammatical structure of a language. For example, nouns, verbs, adjectives, and adverbs make up a sentence or lexical structure in the English language.
Artificial Intelligence (AI): A field in computer science that in working towards creating systems that can simulate human like qualities such as speech recognition systems and self-driving cars.
Recurrent Neural Networks (RNN): Artificial neural networks that use deep learning for sequential data or time-series data are known as recurrent neural networks. They make use of memory to store past data or context to improve future results.
Heuristic Evaluation: An application is assessed using a set of rules by one or more usability experts based on a set of usability criteria or heuristics.
Ease of use (EOU): A usability criteria that determines how easy it is for a user to accomplish a task. For example, is it easy for a user to place an order in a few clicks without having to watch a tutorial or take a tour on how to place an order.
Usability Criteria: A set of quality criteria provided by experts in the field to help guide better design of products and services.
Long Short-Term Memory Network (LSTM): A type of recurrent neural network that takes sequence of data as input to generate insights and predictions, such as real-time speech. It is the most used neural network used in the field of artificial intelligence.
Machine Learning (ML): It is a sub field of artificial intelligence that identifies patterns in human conversations and can be used to create chatbots.
Heuristic Evaluation: An application is assessed using a set of rules by one or more usability experts during a heuristic evaluation.
Artificial Neural Networks (ANN): They are based on the concept of neurons in the brain and are a set of nodes connected together through weighted edges that help in recognizing patterns to develop artificially intelligent systems.
Natural Language Processing (NLP): Applying linguistics, computer science and artificial intelligence to develop systems that can process and analyze natural languages, such as Arabic and English, and be able to understand the content and context of the provided text or speech, to generate insights.
Usefulness: A usability criteria that determines whether the features and tasks provided by the system are useful to the user in getting intended work done.