A text analytics framework for understanding the relationships among host self-description, trust perception and purchase behavior on Airbnb

https://doi.org/10.1016/j.dss.2020.113288Get rights and content

Highlights

  • A deep-learning procedure is proposed to predict guests' trust perception.

  • The readability and perspective taking in self-description can help build trust.

  • Excessive positive sentiment expression would decrease trust perception.

  • More introduction on family, openness, service and travel in self-description would be helpful.

  • The perceived trust derived from self-description does contribute to purchases.

Abstract

Trust plays an important role in sharing transactions on short-term rental platforms. However, the impact of host self-description on trust perception and whether trust perception can influence purchase behavior remain under-studied. Therefore, a text analytics framework was proposed to research the relationships among host self-description, trust perception and purchase behavior on Airbnb. Specifically, a deep-learning-based method was designed to automatically code trust perception of host self-descriptions. And the linguistic and semantic features of description texts were extracted with text mining methods. The estimated order quantity was used to quantify purchase behavior. Then, the influence of linguistic and semantic features on trust perception was identified, and the relationship between trust perception and purchase behavior was also verified. The empirical analysis derives the following findings: i. The readability of self-description is positively associated with trust perception; ii. Perspective taking expressed in self-description is also helpful; iii. Excessive positive sentiment expression can raise barriers to trust building; iv. Paying more attention to family relationship, openness, service and travel experience in self-description would be helpful; v. Trust perception can promote purchases. These findings can help hosts write better self-description, which contributes to trust building and purchases on short-term rental platforms.

Introduction

The rapid development of information technology and network services has brought about the sharing economy boom, and various kinds of online platforms, such as Airbnb, Uber, and Zipcar, have emerged, enabling people to engage in providing services and sharing idle resources with others [1]. Airbnb, a peer-to-peer (P2P) accommodation sharing platform, is one of the most successful models in the sharing economy, providing short-term rental services for more than 200 million guests across 191 countries [2]. On Airbnb, guests make bookings via the online platform and pay for the accommodation service. A complete transaction usually contains the following steps: (1) the guests search for target units on the website and then attempt to contact the hosts via an instant messaging system; (2) the two parties attempt to build trust through their reputation, self-disclosure and communication; (3) once an agreement is reached, guests can check-in and will likely have some face-to-face interactions with the hosts for a period of time; and (4) finally, the guests check out and post a review about the accommodation experience. For the parties engaged in the transaction, hosts can earn money and make friends by sharing idle homes or rooms with guests, and guests can directly interact with hosts and obtain low-cost accommodations. From this perspective, sharing accommodations on Airbnb can be deemed a mixed-mode interaction of social exchange (i.e., online first, then offline) [3]. However, this kind of social exchange may involve higher risks and greater uncertainty. For instance, guests usually begin their social interactions with expectations of gaining a unique accommodation experience, building social relationships, and understanding different cultures [4]. However, these expectations cannot be guaranteed because the online service is intangible, and the quality of the stay cannot be verified until it is experienced [5]. Moreover, guests are also at risk of facing unreliable hosts or even threats to their personal safety.

Confronted with high risk and uncertainty, guests use whatever information they can obtain, such as online reputation and host profile details, to make trust inferences and purchasing decisions. Among these variables, rating scores and reviews, as components of reputation systems, have been widely researched in traditional e-commerce. Consequently, trust-building is usually simplified to the use of a reputation system [6]. However, trust on Airbnb is far more complicated and does not rely solely on reputation. In fact, the comments on Airbnb are often highly positive [7]. Thus, the reputation system alone may not be enough to identify reliable hosts. Faced with the complex scenario of building trust in the sharing economy, hosts are encouraged to make self-disclosures to reduce guests' sense of uncertainty and risk, such as providing a personal photo or writing a self-description. Thus, whether and how these host attributes can impact guests' trust perception and decision-making has attracted researchers' attention.

The existing literature mainly focuses on the impact of a host's numeric attributes, reputation and profile photo on guests' perceived trust and purchase intention. Some studies have found that host gender, emotion, and attractiveness revealed in personal photos influence guests' perceived trust and purchase intention [[8], [9], [10], [11]]. Moreover, the superhost badge, response behavior, and verifications can also impact guests' perceived trust [11]. Self-descriptions can reveal more details about hosts than a profile photo. Specifically, hosts can introduce themselves on their personal page and express any topics and sentiments through various writing styles. The numbers of words and topics contained in self-descriptions were found to have a positive influence on guests' perceived trust [12]. Similarly, positive sentiment being expressed in a host's self-description can also help build guests' perceived trust [11]. However, the types of linguistic style and semantic content of host self-descriptions that can help build guests' perceived trust on Airbnb remain understudied. Moreover, prior studies have often focused on the relationship between guests' trust perception and purchase intention, but little is known about how guests' perceived trust influences purchase behavior.

In this paper, we propose a text analytics framework for understanding the relationships among self-description, trust perception and purchase behavior based on real-world data from Airbnb. A deep-learning procedure was employed to automatically code trust perception. The linguistic features, including readability, sentiment intensity, and perspective taking, were extracted and quantified using text mining methods. The semantic topics hidden in self-descriptions were identified by LDA (latent Dirichlet allocation) [13]. Then, robust regression was applied to identify the impact of linguistic features and semantic features on trust perception. Finally, the positive influence of trust perception on purchase behavior was verified. Here, purchase behavior was quantified by the estimated order quantity. Our study obtained several interesting findings: (1) both readability and perspective taking have a significant positive impact on trust perception, while sentiment intensity has an inverted U-shaped relationship with trust perception; (2) we found that hosts who prefer to express their family relationships, openness, service and travel experience can obtain higher trust perceptions than hosts who mainly express demographic-related information such as age, occupation and personal interests; and (3) we also found that guests' trust perception can promote purchases on Airbnb. These findings can help hosts improve their self-descriptions, which contributes to trust building and purchase behavior on short-term rental platforms.

The remainder of this paper is organized as follows. Section 2 summarizes related studies. In Section 3, the data and methods are introduced. Section 4 presents the results and analysis. In Section 5, the findings, conclusions, implications, and limitations of the work are discussed.

Section snippets

Trust in the sharing economy

The sharing economy is an economic phenomenon that enables the public to obtain income without changing ownership by sharing under-utilized resources [14]. In contrast to traditional e-commerce, the sharing economy pays more attention to providing services such as short-term rental services and ride-hailing services, and the parties involved in sharing transactions usually need to meet with one another face-to-face. Consequently, the quality of these services is usually difficult to guarantee [5

Dataset and method

The text analytics framework employed to understand the relationships among host self-description, trust perception and purchase behavior is shown in Fig. 1. The framework consists of three main parts: data engineering, feature extraction and data analysis. In general, we first collected and cleaned the experimental data. A deep-learning-based method was applied to automatically code the value of trust perception. Then, the linguistic features (e.g., readability, sentiment intensity,

The effect of textual features on trust perception

We performed robust regression to investigate the effect of linguistic features and semantic features on trust perception. The use of robust regression ensures that our model will not be overly affected by violations of underlying assumptions. Compared with widely used methods of regression, such as ordinary least squares, robust regression is less likely to be impaired by outliers, heteroscedasticity and non-normality [50].

The number of notional words of self-description was used to represent

Discussion and conclusion

Trust has been widely considered an urgent problem on Airbnb. However, how the self-description of hosts can influence guests' trust perception and whether trust perception can lead to purchase behavior remain unclear. This paper proposed a text analytics framework for understanding the relationships among self-description, trust perception and purchase behavior. In detail, we proposed a deep-learning procedure to automatically code guests' trust perception. Additionally, the linguistic and

CRediT authorship contribution statement

Le Zhang: Conceptualization, Methodology, Software, Writing - original draft, Validation, Investigation, Resources. Qiang Yan: Conceptualization, Supervision, Project administration, Funding acquisition. Leihan Zhang: Formal analysis, Data curation, Methodology, Software, Writing - review & editing, Visualization.

Acknowledgments

This work was supported by the MOE (Ministry of Education in China) Project of Humanities and Social Sciences [grant number 16YJA630063], the Scientific Research Foundation by MOE and CMCC (China Mobile Communications Corporation) [grant number MCM20170505], National Natural Science Foundation of China [grant number 61671070].

Le Zhang is a Ph.D. candidate in the School of Economics and Management, Beijing University of Posts and Telecommunications, Beijing, China. Her research interests include trust in the sharing economy, social computing, natural language processing.

References (55)

  • N. Hu et al.

    Manipulation of online reviews: an analysis of ratings, readability, and sentiments

    Decis. Support. Syst.

    (2012)
  • S.M. Norman et al.

    The impact of positivity and transparency on trust in leaders and their perceived effectiveness

    Leadersh. Q.

    (2010)
  • Q. Ye et al.

    The impact of online user reviews on hotel room sales

    Int. J. Hosp. Manag.

    (2009)
  • D. Guttentag

    Airbnb: disruptive innovation and the rise of an informal tourism accommodation sector

    Curr. Issue Tour.

    (2015)
  • A. Wilson et al.

    Services Marketing: Integrating Customer Focus Across the Firm

    (2012)
  • G. Zervas et al.

    A first look at online reputation on airbnb, where every stay is above average

  • J. Wu et al.

    In sharing economy we trust: the effects of host attributes on short-term rental purchases

    Int. J. Contemp. Hosp. Manag.

    (2017)
  • X. Ma et al.

    Self-disclosure and perceived trustworthiness of airbnb host profiles

  • D.M. Blei et al.

    Latent dirichlet allocation

    J. Mach. Learn. Res.

    (2003)
  • R. Botsman et al.

    What’s Mine Is Yours: How Collaborative Consumption Is Changing the Way We Live

    (2011)
  • J. Barbalet

    Self-interest and the theory of action

    Br. J. Sociol.

    (2012)
  • J. Bridges et al.

    If nearly all airbnb reviews are positive, does that make them meaningless?

    Curr. Issue Tour.

    (2018)
  • C.R. Berger et al.

    Some explorations in initial interaction and beyond: toward a developmental theory of interpersonal communication

    Hum. Commun. Res.

    (2006)
  • J. Delhey et al.

    How general is trust in “most people”? Solving the radius of trust problem

    Am. Sociol. Rev.

    (2011)
  • D. Christozov et al.

    A measure of risk caused by information asymmetry in e-commerce

    Issues Informing Sci. Inf. Technol.

    (2006)
  • A. Usman et al.

    Exploring the relationship between social media and social influence

  • C. Cheshire

    Online trust, trustworthiness, or assurance?

    Daedalus

    (2011)
  • Cited by (0)

    Le Zhang is a Ph.D. candidate in the School of Economics and Management, Beijing University of Posts and Telecommunications, Beijing, China. Her research interests include trust in the sharing economy, social computing, natural language processing.

    Qiang Yan is a professor in the School of Economics and Management at Beijing University of Posts and Telecommunications. His researches focus on E-commerce and information systems. He has published more than 100 papers in peer-reviewed journals.

    Leihan Zhang is a research associate in the Wangxuan Institute of Computer Technology, Peking University, Beijing, China. His research interests include data mining, complex network, and natural language processing.

    View full text