| With the rapid development of sharing economy,short-term rental has become an indispensable accommodation option for tourists.And short-term rental is the focus point of the sharing economy.Airbnb,as an online market-place,has been providing short-term rental services to global travelers through a peer-to-peer(P2P)model.Hosts and guests exchange information through the platform:hosts publish information about their apartments,houses or bedrooms and potential guests search houses and create booking requests.Once an agreement is reached and the order is paid,guests can check-in and enjoy the accomodation.However,there exists a great trust deficit problem because of the information asymmetry between the hosts and guests.On the one hand,guests are concerned about the authenticity of information provided by hosts.And confronted with inevitable face-to-face interaction or joint living with hosts personal and property safety turns to be additional worries.Therefore,guests are not only faced with property risk but also safety risk.For instance,it is reported that a guest was attacked by his host in Salta,Argentina during an Airbnb short-term rental and was hospitalized for two days.On the other hand,guests cannot perceive the short-term rental service quality in advance.Take the case of Airbnb,nearly 95%of reviews have an average rating higher than 4.5 stars and the reviews tend to comprise a very restricted set of highly positive commentary,which results in that nearly all Airbnb reviews are positive.The slight difference usually makes it difficult for guests to make decisions based solely on the reputation(rating score and reviews).In fact,except for reputation,many other factors can also influence guests’ trust.In this case,how to make full use of host-related information to build the trust and reduce the loss caused by asymmetric information,has become a research focus in the field of short-term rental.Existing researches mainly focus on the influence of few isolated host attributes on guests’ trust by using survey and experimental approaches,and experiment dataset are usually limited and subjective.And a handful of studies is based on objective data of short-term platform,but majority of these work only concentrate on hosts’ numeric attributes.In other words,the various types and the large amount of objective data on the short-term platform,such as text,photo,etc.,are still underutilized.In addition,the existing research mainly focused on the effect of host-related attributes and reputation on guests’ trust,and the dynamic transference of trust among the hosts is still rarely explored.Namely,whether the previous guests’ trust towards the host can impact the following guests’ trust towards the particular host.Considering the research gap,we aim to research the trust influential factors and transference of trust based on the abundant data of Airbnb.We take host’s self-description,profile photo and guest’s review texts as research objects,then use text mining,face recognition and statistical analysis methods to extract and quantify the textual,image and numeric features from host’s self-description,profile photo and guest’s review texts based on the uncertainty reduction theory,the elaboration likelihood model,stereotypes,emotional contagion and trust transfer theory.In this way,we make the empirical research on how host-related attributes impact guests’ trust.Based on the NRC dictionary,we make sentiment analysis on guests’ reviews to explore the trust transference among guests.This paper integrates the methods and theories of management science and computer science.We provide the variables impacting guests’ trust and build research models using the theories in management science.The big data technology in computer science area is applied to analyze the richer and objective data on Airbnb,then extract and quantify the variables from the data.We use the economics models,such as robust regression,dynamic panel regression,etc.,to validate the relationship between provided variables and guest’s trust.Finally,the objective factors affecting guest’s trust are extracted,and the trust transfer among guests is proved.This work effectively makes up for the research gap of existing studies,and provides new research ideas and methods for trust research in sharing economy,which is significance of in guide the trust building in short term rental.The main contribution of this paper are as follows:First,based on the uncertainty reduction theory(URT)and elaboration likelihood model(ELM),we build the theory model to research the effect of textual features of host self-description on guests’trust.In detail,the readability,information amount,sentiment and semantic topics diversity of host self-description are quantified based on the above theories to explore their effects on guests’ trust.Additionally,the effect of different topics hidden in self-description on guests’trust is also researched.From the perspective of social exchange,this paper provides and explains the importance of trust in short-term rental.The sharing accommodation such as Airbnb can be deemed as a kind of social exchange where mixed-mode interactions(i.e.,online first,then offline)occur between hosts and guests.The host-related attributes can directly impact the guest’s trust.Among host-related attributes,the host self-description is an important self-presentation.Aiming at reducing guests’ uncertainty and improving their trust,hosts can introduce themselves by writing self-descriptions to provide personal information for guests.In detail,based on URT and ELM,we propose hypothesis that the readability,information amount,sentiment and topic diversity can influence guests’ trust.We validate these hypothesis by using the robust regression model.In addition,the LDA is used to identify the topics hidden in host self-description.And t test is also applied to analyze the effect of different topics hidden in self-description on guests’ trust.We found that(1)the higher readability of host self-description has a better persuasive effect and can get more guests’ trust.(2)the information amount and topic diversity of host self-descriptions can reduce the guests’uncertainty and improve trust.(3)hosts who express their positive emotions such as enthusiasm,friendly,etc.,can obtain more guests trust.(4)hosts who prefer to express topics concerning travel experience,service attitude and ability,views and objects about sharing accommodation can get more guests’ trust than those who mainly focus on their age,professional and hobby.Second,based on the beauty,facial emotions and stereotypes theories,we build the theory model to explore the effect of host’s facial features in profile photos on guests’trust.In detail,the effects of facial attractiveness,smile intensity,gender,and race exposed from host photos on guests’ trust are investigated.We propose hypothesis that the information of facial attractiveness,smile intensity,gender and race exposed from host profile photo can influence guests’ trust based on beauty,facial emotion,stereotypes theories.And we also make an empirical validation on these hypotheses.In detail,we crawl tens of thousands of host photos from Airbnb.The face++ API is applied to detect and extract facial features,and we make json parsing for the detection results.Finally,the robust regression model is used to analyze the relationship between facial features and guests’trust.We found that(1)facial attractiveness has a significant influence on guests’ trust.The reason is that hosts with higher facial attractiveness tend to charge a higher house price,which improve the cost and risk of guests.(2)the smile intensity in profile photos has an inverted U-shaped relationship with guests’trust.(3)the short-term rental platforms are also sensitive to gender and racial discrimination.Female and white hosts are more likely to get the guests’ trust than male and black hosts.It is perhaps because that the female and the white are usually believed to be having higher enthusiasm and ability.For short-term rental service,the enthusiasm and ability of hosts are important for guests,which can directly determine the accomodation experience and further influence the choice of guests.Third,we build the influential factors theory model of guests’trust based on multi-dimensional attributes of hosts.By R2 decomposition,we explore the effect of each isolated feature and each group of features on guests’ trust.In addition,we apply the deep neural network and integrated learning method to train a prediction model for predicting guests’ trust.This paper focuses on the effect of multi-dimensional attributes of hosts on guests’ trust,and proposes a computational framework for understanding antecedents of guests’ perceived trust.The multi-dimensional attributes of hosts include numeric features(reputation,validation,response behavior),textual features(information quality,sentiment,semantic topics),and image features(facial emotion).The regression model is used to explore the relationship between attributes and guests’trust.The effects of each attribute and group attributes on guests’trust are compared by R2 decomposition.And the prediction model of guests’ trust is built by using the deep neural network and integrated learning methods.We found that reputation plays the most important role in affecting trust building and the contribution of reputation is not less than the contribution of the rest antecedents.The "superhost" badge and reviews have significant positive influence on guests’ trust.However,the review rating has no significant effect on guests’ trust.Hosts who make a positive expression when writing self-description can get more guests’ trust.And hosts who focus on introducing their communication,service ability,etc.,can obtain more guests’ trust.In addition,hosts with profile photo expressing more positive emotions can get more guests’ trust.Finally,the prediction model reach 72.32%,71.56%and 71.56%in precision,F-score and AUC.The results of this research can be used for the design of short-term rental recommendation method in the short-term rental platform.Fourth,based on the trust transfer theory and emotion contagion,we make a research on the trust and emotion transfer among guests on short-term rental.And we also analyze the influence of trust and emotion transfer on guests’ trust.This paper proposed the theory hypothesis that there exist trust and emotion transfer among guests on short-term rental according to the trust transfer theory and emotion contagion.By using the emotion analysis method based on emotional lexicon,we made the sentiment analysis on the previous guests’ reviews to identify and quantify the trust and the other seven emotions embodied in the reviews.The seven emotions are respectively anger,anticipation,disgust,fear,joy,sadness and surprise.Then,the dynamic panel regression analysis was applied to analyze whether the trust and seven emotions embodied in the reviews of the previous group of guests can impact the next group of guests’ trust and emotions.In this way,we can validate that whether the trust and emotion transfer occurs among guests.In addition,we also validate whether the trust and seven emotions expressed by the previous guests can influence the following guests’ trust.We found that(1)the trust expressed in the previous guests has a significant positive influence on the following guests’ trust.And the seven emotions embodied in the previous guests also have a significant positive influence on the following guests’ emotions.The results show that the trust and emotion transfer can occur between guests.We also found that trust and negative emotion have stronger transitivity than other positive emotions.(2)the influence of trust and seven emotions on trust transfer are different.The trust expressed in the previous guests has the most positive influence on the trust transfer.As for the seven emotions,the joy emotion has a significant positive influence on trust transfer,and the anticipation,surprise,disgust have a significant negative impact on trust transfer.But the negative emotions of sadness and fear have a significant positive influence on trust transfer.The reason is perhaps that both negative emotions of sadness and fear and positive emotions exist in the reviews,which can improve the guests’ perceptions of objective and usefulness.We can make the validation in the future work. |
PDF Full Text Request