Navigational apps are relied upon by millions of people every day. However, in addition to helping people navigate the world, many now expect their phones to help them discover new places, too. But is it really possible to wander around exploring a new city and, without any prior planning, locate points of interest (POIs, e.g. museums and restaurants) that interest you?
Apps like Google Maps, Tripadvisor and Instagram are often relied upon by those looking for POI recommendations. However, these apps don’t offer personalized results. Attempting an on-the-go search for POIs that suit personal preferences and current needs can be a frustrating and often lengthy process. An app that can understand users’ unique taste, as well as their current mood, could automatically provide a range of personal and context-specific recommendations.
Our aim is to inform the design of this kind of ‘ambient wandering’ technology. To this end, we’re developing an intelligent contextualized and personalized recommendation system with a focus on delivering POI information to people on the go, based on their location and mindset. Before beginning work on any POI recommendation algorithms, we decided to first research and understand the expectations of end users. For this, we created the concept of an intelligent POI recommending assistant for mobile urban exploration that we named Ambient Wanderer.
We originally envisioned Ambient Wanderer to be used by everyone, in much the same way as a navigational app. When we looked at studies on urban exploration and information retrieval, however, we noticed that there’s been very little focus on newcomers to a city. So, in a spectrum of urban explorers—ranging from tourists who have limited knowledge of the city they’re visiting to local residents who know it pretty well—we chose to recruit people who had recently relocated to a new city (from within their own country or abroad). The POI needs of these newcomers tend to correspond to those of both locals and tourists. Also, prior research has shown that these people are likely to explore by walking, wandering and wayfinding to get familiar with their new habitat . To better characterise their in-between nature, we refer to these people as ‘new locals’. To validate and strengthen our concept, we elicited the feedback of potential end users. Then, using this information, we derived design implications for the development of the system.
In early evaluations by other groups looking to integrate user feedback into intelligent systems, the use of low-fidelity (pen-and-paper) prototypes [2, 3] and concept testing with storyboarding  has helped in the development of better algorithms and features. However, in our case, such low-fidelity approaches are too far removed from the real-life application (i.e. actually using Ambient Wanderer in the street). To test the personalized aspects of Ambient Wanderer, we needed to create profiles for each participant based on their personal information . With these requirements in mind, we customized a hybrid Wizard of Oz prototyping methodology . In a Wizard of Oz experiment, participants believe that the system they are interacting with is autonomous, when it is in fact being operated (partially or wholly) by an unseen human being. Our hybrid approach uses a high-fidelity prototype of Ambient Wanderer, annotated with the participants’ personal preferences and the contextual variables of their environment. By inputting recommended POIs manually, based on a user’s profile (collected from a pre-questionnaire) and the user’s current context (location, weather and time of the day), we were able to remove the human ‘wizard’ and have our prototype effectively fake an ideal recommendation system that supports serendipitous urban exploration.
The user interface of the Ambient Wanderer prototype consisted of three key features: a localized map; POI information; and seven ‘mindset’ options (shown in Figure 1). These mindsets include a range of categories—such as ‘I’m hungry’ and ‘Hidden gems’—that were conceived by our research team during a card-sorting session  to guide the selection of recommended POIs. Unlike traditional POI categories that are descriptions of places (e.g. ‘Restaurants’ and ‘Museums’), mindsets enable users to explore POIs based on a description of their state of mind. In addition to acting as visual cues that provide ideas for on-the-spot exploration of the city, we hoped that the mindsets would also push the participants to think about the POI needs they have that aren’t being catered for within the prototype.
To test our prototype, we enlisted 12 new locals from Grenoble, France. We used critical incident interviews  and asked the participants to use Ambient Wanderer in the field to derive implications for the design of a POI recommendation system that would serve it.
From a qualitative thematic analysis of the experimental sessions with the Ambient Wanderer app, we obtained the following findings regarding user behaviour:
We also succeeded in validating our concept of Ambient Wanderer. The ‘state of mind’ nature of the search, supported by mindsets, was appreciated by the participants (see Figure 2, left). They found our interface ‘clean’ and ‘easy’ and 100% of participants agreed that Ambient Wanderer was fun and engaging. Based on findings from this experiment, we plan to grow our collection of mindsets to provide a wider range of options for a more accurate reflection of user needs. To this end, we also intend to develop customizable and user-generated mindsets and to work on a system that can automate the classification of different POIs into suitable mindsets.
As we’d expected, the results from our study show that the behaviours and needs of new locals are similar to those of locals and tourists. We’re currently comparing our observations to further clarify the global picture of our POI recommendation system. Beyond the development of Ambient Wanderer, we aim to perform studies with a larger sample—including the entire spectrum of locals, new locals and tourists—to assess how well the system fits to the specific needs of each user category.
For a detailed description of how we designed our methodology for the hybrid Wizard of Oz experiment, how we uncovered the urban exploration needs of the new locals and the corresponding implications for designing a POI recommender system, see our full paper at the Designing Interactive Systems 2020 conference .
 Making the City My Own: Uses and Practices of Mobile Location Technologies for Exploration of a New City. Louise Barkhuus and Donghee Yvette Wohn. Personal and Ubiquitous Computing, vol. 23, no. 2, 2019, pp. 269–278. DOI: 10.1007/s00779-018-01191-z.
 Toward Harnessing User Feedback for Machine Learning. Simone Stumpf, Vidya Rajaram, Lida Li, Margaret Burnett, Thomas Dietterich, Erin Sullivan, Russell Drummond and Jonathan Herlocker. Proceedings of the 12th International Conference on Intelligent User Interfaces (IUI ’07), Honolulu, HI, 28–31 January 2007, pp. 82–91. DOI: 10.1145/1216295.1216316.
 Integrating Rich User Feedback Into Intelligent User Interfaces. Simone Stumpf, Erin Sullivan, Erin Fitzhenry, Ian Oberst, Weng-Keen Wong and Margaret Burnett. Proceedings of the 13th International Conference on Intelligent User Interfaces (IUI ’08), Gran Canaria, Spain, 13–16 January 2008, pp. 50–59. DOI: 10.1145/1378773.1378781.
 Triptech: A Method for Evaluating Early Design Concepts. Julie Anne Séguin, Alec Scharff and Kyle Pedersen. Extended Abstracts of the 2019 CHI Conference on Human Factors in Computing Systems (CHI EA ’19), Glasgow, Scotland, 4–9 May 2019, pp. 1–8. DOI: 10.1145/3290607.3299061.
 Wizard of Oz Prototyping for Machine Learning Experiences. Jacob T. Browne. Extended Abstracts of the 2019 CHI Conference on Human Factors in Computing Systems (CHI EA ’19), Glasgow, Scotland, 4–9 May 2019, pp. 1–6. DOI: 10.1145/3290607.3312877.
 Hybrid Wizard of Oz: Concept Testing a Recommender System. Sruthi Viswanathan, Behrooz Omidvar-Tehrani, Adrien Bruyat, Frédéric Roulland and Antonietta Maria Grasso. Extended Abstracts of the 2020 CHI Conference on Human Factors in Computing Systems (CHI EA ’20), Honolulu, HI, April 2020, pp. 1–7. https://doi.org/10.1145/3334480.3383097
 The Critical Incident Technique in Service Research. Dwayne D. Gremler. Journal of Service Research, vol. 7, no. 1, 2004, pp. 65–89. DOI: 10.1177/1094670504266138.
 Designing Ambient Wanderer: Mobile Recommendations for Urban Exploration. Sruthi Viswanathan, Behrooz Omidvar-Tehrani, Adrien Bruyat, Frédéric Roulland and Antonietta Grasso. Proceedings of the 2020 ACM Conference on Designing Interactive Systems (DIS ’20), 6–11 July 2020. DOI: https://dl.acm.org/doi/abs/10.1145/3357236.3395518
You may choose which kind of cookies you allow when visiting this website. Click on "Save cookie settings" to apply your choice.
FunctionalThis website uses functional cookies which are required for the search function to work and to apply for jobs and internships.
AnalyticalOur website uses analytical cookies to make it possible to analyse our website and optimize its usability.
Social mediaOur website places social media cookies to show YouTube and Vimeo videos. Cookies placed by these sites may track your personal data.
This content is currently blocked. To view the content please either 'Accept social media cookies' or 'Accept all cookies'.
For more information on cookies see our privacy notice.