Users’ Choices About Hotel Booking: Cues for Personalizing the Presentation of Recommendations Catalin-Mihai Barbu Jürgen Ziegler University of Duisburg-Essen University of Duisburg-Essen Duisburg, Germany Duisburg, Germany catalin.barbu@uni-due.de juergen.ziegler@uni-due.de ABSTRACT EXPLORATORY STUDY Personalization in recommender systems has typically been We conducted an exploratory online study to investigate applied to the underlying algorithms. In contrast, the participants’ choices about hotel booking. In selecting the presentation of individual recommendations—specifically, domain, we considered three aspects: 1) The choice should the various ways in which it can be adapted to suit the carry a substantial amount of risk for the user; 2) the items user’s needs in a more effective manner—has received should have a reasonable set of attributes that need to be relatively little attention by comparison. We present the considered; and 3) there should be a large body of user- results of an exploratory survey about users’ choices generated content available, in the form of reviews, photos, regarding hotel recommendations and draw preliminary tags, and ratings, that can be leveraged for the presentation. conclusions about whether these choices can influence the Because of the first criterion, we decided against using the presentation of recommendations. more common domain of movie recommendations. Author Keywords Study Design Recommender systems; personalization; user study; tourism We theorize that the way in which people make decisions about hotel booking, their trust in social media, and their ACM Classification Keywords H.5.2 [Information Interfaces and Presentation]: User travel habits influence the information they want to see in a Interfaces—evaluation/methodology, graphical user recommendation (i.e. the type of personalization they interfaces (GUI), user-centered. expect). Our aim for this study was to investigate whether the travel scenario influences users’ decision-making INTRODUCTION & MOTIVATION processes in ways that can be used to personalize the Personalization is an important and well-studied topic in presentation of hotel recommendations. recommender systems (RS). Previous research has noted the positive effect of personalization on enhancing user The survey was organized in six parts. The first four experience [5]. A relatively unexplored area concerns the sections elicited answers regarding our participants’ personalization of the presentation of recommendations. demographics, trust in social media, experience with hotel Elicited user preferences can be used not only to offer booking portals, and travel behavior. A filter question was personalized predictions, but also to customize the way in used to assign each participant to one of five travel which these predictions are presented to the user. Adapting scenarios: city break / short vacation (1-2 nights), short the presentation to fit individual needs has the potential to business trip (1-2 nights), long vacation (3+ nights), long uncover novel interaction possibilities. We present the business trip (3+ nights), or family vacation (with children). results of an exploratory study that investigated users’ In each scenario, users were presented with an identical choices about the presentation of hotel recommendations mockup of a hotel recommendation. First, participants were and our preliminary conclusions on whether these choices asked to rank each section of the mockup—overall rating, could influence the presentation of recommendations. price, general description of the hotel, photos, a map RELATED WORK showing the hotel’s location within the city, nearby Some of the main research foci of personalization include transportation options, hotel and room amenities, and deciding, for a given recommendation, what information to reviews from users—depending on how important they present, when to present it [1], how much of it to present considered the information in that section to be. Second, [2], and in what way [6]. Many existing approaches to they had to select up to 7 topics about which they would personalizing the presentation of recommendations rely on like to receive more information when looking at explanations (see, e.g., [7] for an in-depth analysis of the recommendations (e.g., proximity to public transport, room effects). So-called “common sense” approaches, which use sizes and layouts, or fitness center equipment). rules to determine what items to recommend and how to Finally, participants were asked 12 questions designed to personalize the presentation have also been developed [3]. determine their typical decision-making behavior during hotel booking. This section was modelled based on the Copyright is held by the author(s). Rational-Experiential Inventory [4], which is designed to RecSys 2017 Poster Proceedings, August 27-31, Como, Italy. RecSys 2017 Poster Proceedings, August 27-31, Como, Italy Barbu and Ziegler measure participants’ need for cognition and faith in intuition, respectively. The questions addressed six underlying factors: a) perceived effort required to complete a hotel booking task; b) economic considerations; c) clearness of mental goal; d) self-efficacy (i.e. trust in one’s own choices); e) influenceability; and f) engagement. Two questions (high and low loading) were asked per factor. Study Results The survey was published online in January 2017 and ran for one month. A total of 159 participants (82 female; median age in the interval 25-34 years) completed the survey fully. Of the respondents, 123 (77.36%) were employed and 24 (15.09%) were students. Furthermore, 139 (87,42%) had completed at least a Bachelor education. As monetary incentive, all complete responses entered a raffle for one of four Amazon gift vouchers, each worth 25 EUR. Most participants (51%) rated their trust in online reviews as high or very high on a 5-point Likert scale (M=3.53, SD=0.71). These findings were similar across all scenarios. After data analysis (ANOVAs with Fisher’s LSD), we Figure 1: Results of users’ decision-making behavior during noticed a significant difference (p < 0.05) when comparing hotel booking. Error bars denote the 95% confidence interval. the business scenarios: Over 65% of participants whose typical travel scenario was “long business trip” reported a Initial findings suggest that the motivation behind searching high or very high trust in online reviews, compared to only for a recommendation influences users’ decision processes. 48% in the “short business trip”. The availability of reviews As ongoing work, we are investigating potential links was rated as very or extremely useful by 78% of between individual factors and presentation preferences. participants (M=3.96, SD=0.75). Similarly, photos were ACKNOWLEDGMENTS considered very or extremely useful by 82% of respondents This work is supported by the German Research Foundation (M=4.17, SD=0.82). In both cases, we observed no (DFG) under grant No. GRK 2167, Research Training significant differences between travel scenarios. Group "User-Centred Social Media". Certain patterns emerged with respect to users’ typical REFERENCES decision-making behavior during hotel booking. First, 1. Betzalel, N. D., Shapira, B., & Rokach, L. “Please, not now!” booking a hotel for vacation is considered more challenging A model for timing recommendations. In Proc. RecSys '15, than for business travel—especially for longer stays ACM (2015), 297–300. (p < 0.05). Second, people who typically go on longer 2. Bollen, D., Knijnenburg, B. P., Willemsen, M. C., & vacations need more time to decide which recommendation Graus, M. Understanding choice overload in recommender to follow when prices are higher than they are used to. The systems. In Proc. RecSys '10, ACM (2010), 63–70. difference was significant (p < 0.05) when compared to the 3. De Carolis, B., Mazzotta, I., Novielli, N., & Silvestri, V. answers from the “long business trip” scenario. Third, Using common sense in providing personalized participants tend to revisit recommendations to ensure they recommendations in the tourism domain. In CARS '09. do not miss important information. A significant difference 4. Epstein, S., Pacini, R., Denes-Raj, V., & Heier, H. Individual (p < 0.01) was observed when comparing the scenarios differences in intuitive–experiential and analytical–rational “short vacation” and “short business”. These results suggest thinking styles. Journal of personality and social that the travel scenario can be a factor for personalizing the psychology 71-2, (1996). presentation of recommendations. However, its influence 5. Knijnenburg, B. P., Willemsen, M. C., Gantner, Z., Soncu, may be lower than predicted (Figure 1). H., & Newell. C. Explaining the user experience of Contrary to our expectations, we observed almost no recommender systems. User Mod. User-Adap. 22, 4-5 significant differences in terms of the importance of the (2012), 441–504. mockup sections for the different scenarios. The sole 6. Nanou, T., Lekakos, G., & Fouskas, K. The effects of exception (p < 0.01) was “general hotel description”, which recommendations’ presentation on persuasion and proved particularly unimportant for respondents in the satisfaction in a movie recommender system. Multimedia “long business” scenario. Similarly, the list of topics about Systems 16, 4-5 (2010), 219–230. which participants stated they would like to see more 7. Tintarev, N., & Masthoff, J. Designing and evaluating information when browsing recommendations did not explanations for recommender systems. Recommender exhibit significant differences across scenarios. Systems Handbook. Springer US (2011), 479–510.