Keynote - Wolfgang Wörndl, TU Munich, Germany

Utilizing Location-based Social Media for Trip Mining and Recommendation

Delivering personalized and timely information is particularly valuable in mobile scenarios such as traveling users. Therefore it is desirable to assist the user by services that are tailored towards her context, e.g. the location. Recommender systems recommend movies, restaurants or other items to an active user based on information about users and items, and can be an effective method to support travelers. However, this domain poses special challenges such as high risk, data sparseness, cold start and the complexity of combining items [1]. This talk will first present a data-driven method to mine trips from location-based social networks (LBSN) to better understand traveler mobility [2]. These trips can be quantified using a number of metrics to capture the underlying travel patterns. We outline two use cases to utilize the mined trips: clustering travelers and also destinations.
In the second part, we describe an approach to generate routes consisting of sequences of points-of-interests (POIs) for city walking tours [3]. We discover and score POIs based on retrieving places from the LSBN Foursquare, so it is not restricted to certain cities or regions. Discovered places are then combined to a practical route using Dijkstra’s shortest path algorithm. We show how to find trips that maximizing the value for a user while respecting time and budget constraints. The solution has been implemented in a practical web and mobile application. Since people often travel in groups, we have also looked into different strategies that can be used to recommend POI sequences to groups [4]. We propose a novel strategy called Split Group, which allows groups to split into smaller groups during a trip. The methodology can be generalized to recommend other complex composite items. We conclude the talk with a summary and an outline of current and future challenges for tourism recommender systems which includes multistakeholder aspects and fairness issues.

References
[1] Herzog, D., Dietz, L. W., & Wörndl, W. (2019). Tourist trip recommendations—foundations, state of the art and challenges. Personalized Human-Computer Interaction, 6, 159-182.
[2] Dietz, L. W., Sen, A., Roy, R., & Wörndl, W. (2020). Mining trips from location-based social networks for clustering travelers and destinations. Information Technology & Tourism, 22(1), 131-166.
[3] Wörndl, W., Hefele, A., & Herzog, D. (2017). Recommending a sequence of interesting places for tourist trips. Information Technology & Tourism, 17(1), 31-54.
[4] Herzog, D., & Wörndl, W. (2019, September). User-centered evaluation of strategies for recommending sequences of points of interest to groups. In Proceedings of the 13th ACM Conference on Recommender Systems (pp. 96-100).

Wolfgang Wörndl is a senior researcher and lecturer at the School of Computation, Information and Technology (CIT) at Technische Universität München (TUM). His research focuses on human-centered and interactive recommender systems in mobile scenarios such as travel and tourism. He has published over 100 refereed papers in related areas. He was co-organizing and served as program committee member for a large number of journals, conferences and workshops, including the ACM RecSys workshop series on Recommenders in Tourism.