Review article
Conversational agents in business: A systematic literature review and future research directions

https://doi.org/10.1016/j.cosrev.2020.100239Get rights and content

Highlights

  • The review encompasses literature of the last decade through a well-defined methodology.

  • Three general, three focused, and two statistical questions are addressed.

  • Aspects of artificial intelligence of conversational agents in business are reviewed.

  • No study combines self-learning, personalization, and generative-based responses.

  • Opportunities towards human-like conversational agents in business are presented.

Abstract

The field of business shows an increasing interest in exploring conversational agents to improve service quality and market competitiveness. Furthermore, the advances in machine learning capabilities leverage the natural language processing towards natural and straightforward dialogue experiences for industries. However, in the best of our knowledge, no literature review outlines conversational agents in the business industry, primarily taking into account computational learning capabilities. This article presents a systematic literature review that encompasses these areas looking through the use of machine learning to improve the field of business. The review followed a guideline for systematic reviews to present the literature of the last decade, emphasizing business perspectives such as domains, goals, and challenges, and computational methods for self-learning, personalization, and response generation of conversational agents. As a result, the article provides the answers of three general, three focused, and two statistical questions to address the role of artificial intelligence in conversational agents applied to business domains. In this regard, the results show that no study combines self-learning, personalization, and generative-based responses for the same business solution. Additionally, the article describes the organization of the state-of-the-art, highlighting the correlation of business perspectives and machine learning methods. The contributions of this review focus on opportunities and future research directions towards human-like conversational agents for business.

Introduction

Technological investments in the field of business have shown an increasing interest in exploring conversational agents [1]. The estimates indicate that the application of these agents will be responsible for cost savings of $11.5 billion by 2023 [2]. These gains originate from the highly available, scalable, and cost-effective services provided by conversational agents [3], focusing on the better usage of technology [4] and improving service quality and market competitiveness [5]. Furthermore, these services increase user engagement by offering customized flexibility and experiences that are simple and natural [6]. However, like any other dialogue system, limitations arise from the grammatical complexity or semantics of the conversation, such as inappropriate responses generation [7].

In recent years, advances in artificial intelligence methods enabled an upsurge of powerful technologies, for example, deep neural networks. Hence, several works introduced these techniques in the dialogue pipeline, usually composed by Natural Language Understanding (NLU), Dialogue State Tracking (DST) and Natural Language Generation (NLG) [8]. Additionally, end-to-end frameworks became popular in both non-task-oriented and task-oriented dialogue systems [9]. Such methods, in combination with strategies of digital transformation, contributed to current research perspectives [10].

In addition, the state-of-the-art encompasses a more specific set of topics, for instance, self-learning and personalization methods. In this regard, some recent reviews focused on conversational agents from a technical perspective [9], [11], [12], [13]. At the same time, emerged business opportunities require more capable conversational agents (e.g., provided with human-like features) which have been considered by others reviews [14], [15], [16]. However, in the best of our knowledge, no literature review outlines the conversational agents in business domains, primarily taking into account machine learning methods.

This article addresses this gap with a systematic literature review that explores the following question: what is the role of artificial intelligence in the last decade on the state-of-the-art of conversational agents applied to business domains? Therefore, we conducted a literature review with a well-defined methodology to answer three general, three focused, and two statistical research questions. The contributions of this literature review are: (i) provide a mapping of the business domains explored since last decade, describing computational methods for self-learning, personalization and response generation of conversational agents; (ii) outline the state-of-the-art focusing on the business support through conversational agents; and, (iii) identify opportunities and future research directions towards more capable conversational agents in business.

The remainder of this article is organized as follows. Section 2 describes the related works. Section 3 details the materials and methods used in this literature review. Section 4 presents the results with a focus on the research questions. Section 5 discusses the limitations of this research. Section 6 approaches a discussion of the findings, followed by Section 7, which presents the opportunities and future research directions over the field. Finally, Section 8 provides the final considerations.

Section snippets

Related work

The aim of this study is to review the literature of conversational agents in the business domains with a focus on machine learning. In this regard, some surveys have been conducted to review conversational agents from a technical perspective.

The study proposed by Hussain et al. [11] focuses on chatbot classification and chatbot design techniques, where the authors explored task-oriented and non-task oriented categories of chatbot. These categories were taken into account in a discussion on how

Materials and methods

This section presents the methodology used to conduct the study. It follows the principles of systematic reviews, as proposed by Petersen et al. [17] and updated by Petersen, Vakkalanka, and Kuzniarz [18]. The methodology encompasses the planning and execution of the literature review, and employs five well-defined steps:

  • 1.

    Definition of research questions

  • 2.

    Definition of search processes

  • 3.

    Definition of the criteria for article selection

  • 4.

    Execution of data extraction

  • 5.

    Execution of analysis and

Results

This section presents the results of the literature review, structured according to the research questions. As suggested by Dalmina et al. [77] and Dias et al. [78], the previously described systematic review methodology helps mitigate possible biases in the results. First, this section outlines the methodology execution, associating each step with the number of selected articles. Then, it describes the main findings by answering each of the research questions.

Fig. 1 illustrates the filtering

Limitations

As any other literature review, this study presents limitations that may affect the scope of the results. However, the decisions taken during the planning and execution try to mitigate them.

In the planning of this review, six research databases were selected to reduce bias. These sources contain peer-reviewed publications from Computer Science and Information Technology. We manually defined the search string, considering two major terms and search terms. The search terms were selected

Discussion

This section presents a discussion of the findings based on the analysis of the results. The findings outline the literature of the last decade and illustrate the business perspectives that have been supported by conversational agents.

In the analysis of the studies to answer the general questions, we noticed the exploration of nine different business domains during the last decade. However, the majority of them focus on commerce, more specifically, e-commerce. This result may be due to the

Opportunities and future directions

Transparent and human-like conversational agents are determinant factors of the users’ trust during the dialogue [80]. One promising step towards these features should combine the three methods of the focused questions from this review, i.e., self-learning, personalization, and generative-based techniques. The self-learning allows to retrieve useful knowledge and proper features of users’ situation during the interaction. This knowledge should be preserved and maintained as historical

Conclusion

This article presented a systematic literature review about the use of conversational agents in business domains. The review answered three general, three focused, and two statistical research question based on a selected literature corpus. The corpus was created through the search of articles in research databases, with inclusion and exclusion criteria followed by filtering steps. Each article of the corpus was used to answer the proposed research questions, presenting the role of artificial

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgements

The authors would like to thank Dell Inc. for funding this work. We are also grateful to University of Vale do Rio dos Sinos (UNISINOS) and Applied Computing Graduate Program (PPGCA) for embracing this research.

References (83)

  • ChungM. et al.

    Chatbot e-service and customer satisfaction regarding luxury brands

    J. Bus. Res.

    (2018)
  • ChenH. et al.

    A survey on dialogue systems: Recent advances and new frontiers

    SIGKDD Explor. Newsl.

    (2017)
  • MaedcheA. et al.

    AI-Based digital assistants

    Bus. Inf. Syst. Eng.

    (2019)
  • HussainS. et al.

    A survey on conversational agents/chatbots classification and design techniques

  • LokmanA.S. et al.

    Modern chatbot systems: A technical review

  • NuruzzamanM. et al.

    A survey on chatbot implementation in customer service industry through deep neural networks

  • KaghyanS. et al.

    Review of interactive communication systems for business-to-business (B2B) services

    Electron. Imaging

    (2018)
  • Meyer von WolffR. et al.

    How may i help you? – state of the art and open research questions for chatbots at the digital workplace

  • PfeufferN. et al.

    Anthropomorphic information systems

    Bus. Inf. Syst. Eng.

    (2019)
  • PetersenK. et al.

    Systematic mapping studies in software engineering

  • KeshavS.

    How to read a paper

    ACM SIGCOMM Comput. Commun. Rev.

    (2007)
  • SubramaniamS. et al.

    COBOTS - A cognitive multi-bot conversational framework for technical support

  • SinghM. et al.

    KNADIA: Enterprise knowledge assisted dialogue systems using deep learning

  • QiuM. et al.

    Transfer learning for context-aware question matching in information-seeking conversations in e-commerce

  • XuA. et al.

    A new chatbot for customer service on social media

  • SunY. et al.

    Conversational recommender system

  • CuiL. et al.

    Superagent: A customer service chatbot for e-commerce websites

  • ToxtliC. et al.

    Understanding chatbot-mediated task management

  • NiculescuA.I. et al.

    SARA: Singapore’s automated responsive assistant, a multimodal dialogue system for touristic information

  • ChenS. et al.

    Review-driven answer generation for product-related questions in e-commerce

  • MukherjeeS. et al.

    Help yourself: A virtual self-assist system

  • XueZ. et al.

    Isa: Intuit smart agent, a neural-based agent-assist chatbot

  • YanZ. et al.

    Building task-oriented dialogue systems for online shopping

  • ChandarP. et al.

    Leveraging conversational systems to assists new hires during onboarding

  • ZhuP. et al.

    Lingke: a fine-grained multi-turn chatbot for customer service

  • DeksneD. et al.

    Collection of resources and evaluation of customer support chatbot, vol. 307

    (2018)
  • ZhaoR. et al.

    SOGO: A social intelligent negotiation dialogue system

  • GriolD. et al.

    A multiagent-based technique for dialog management in conversational interfaces

  • AtiyahA. et al.

    An efficient search for context-based chatbots

  • LommatzschA.

    A next generation chatbot-framework for the public administration

  • DündarE.B. et al.

    A hybrid approach to question-answering for a banking chatbot on turkish: Extending keywords with embedding vectors

  • Cited by (84)

    • Machine learning-based automation of accounting services: An exploratory case study

      2023, International Journal of Accounting Information Systems
    View all citing articles on Scopus
    View full text