Deciding where to travel is a complex, emotionally involving, and financially relevant decision which people face relatively infrequently. Some aspects of tourist recommendations, such as point-of-interest recommendation, hotel recommendations, or restaurant recommendations, are commercially well established, whereas there are few successful recommender systems for individual travel destinations. In this thesis, we present several contributions in the context of destination recommendation covering traveler mobility analysis, destination characterization, and conversational recommender systems.
Understanding traveler mobility forms the basis for more personalized recommendations. We propose methods to analyze global traveler mobility from location-based social networks to learn which data sources are suitable for analyses in this domain, how people travel around the world, and which types of travelers can be observed. Our cluster analyses of trips and travelers reveal distinct groups, which can serve as an initial preference elicitation step, but we could also show that the common practice of evaluating point-of-interest recommendations without differentiating these groups leads to misleading results. Furthermore, we use the mined trips to construct a specialized map of hierarchical travel regions and to recommend the personalized duration of stay at destinations.
To correctly match traveler preferences with destinations in the content-based recommendation domain, we investigate which data sources are suitable for characterizing destinations. Constructing 18 data models and eliciting the concept of touristic experience in cities using an expert study, we determine that textual data sources, e.g., Wikipedia articles, do a good job of emulating the touristic experience using rank agreement metrics. Additionally, we are able to optimize data sources with explicit features to be competitive by learning the importance of feature weights using black-box learning.
Finally, we present the CityRec conversational destination recommender system. Since users often struggle to verbalize their true preferences and might have unrealistic expectations about destinations, we propose a novel conversational paradigm, “Navigation by Revealing Trade-offs”, to overcome the wishful-thinking problem and inform users of the trade-offs involved in choosing one destination over another. The seamless integration of user interface and algorithms was evaluated using a large-scale user study with 600 participants.
Linus W. Dietz. “Data-Driven Destination Recommender Systems.” Technical University of Munich, 2023.
]]>The involvement of geographic information differentiates point-of-interest recommendation from traditional product recommendation. This geographic influence is usually manifested in the effect of users tending toward visiting nearby locations, but further mobility patterns can be used to model different groups of users. In this study, we characterize the check-in behavior of local and traveling users in a global Foursquare check-in data set. Based on the features that capture the mobility and preferences of the users, we obtain representative groups of travelers and locals through an independent cluster analysis. Interestingly, for locals, the mobility features analyzed in this work seem to aggravate the cluster quality, whereas these signals are fundamental in defining the traveler clusters. To measure the effect of such a cluster analysis when categorizing users, we compare the performance of a set of recommendation algorithms, first on all users and then on each user group separately in terms of ranking accuracy, novelty, and diversity. Our results on the Foursquare data set of 139,270 users in five cities show that locals, despite being the most numerous groups of users, tend to obtain lower values than the travelers in terms of ranking accuracy while they also seem to receive more novel and diverse POI recommendations. For travelers, we observe the advantages of popularity-based recommendation algorithms in terms of ranking accuracy by recommending venues related to transportation and large commercial establishments. However, there are considerable differences in the respective traveler groups, especially between predominantly domestic and international travelers. Due to the considerable influence of mobility on the recommendations, this article underlines the importance of analyzing user groups differently when making and evaluating personalized point-of-interest recommendations.
Pablo Sánchez and Linus W. Dietz. “Travelers vs. Locals: The Effect of Cluster Analysis in Point-of-Interest Recommendation.” In: 30th ACM Conference on User Modeling, Adaptation and Personalization. UMAP’22. New York, NY, USA: ACM, July 2022, pp. 132– 142. DOI: 10.1145/3503252.3531320
]]>Characterizing items for content-based recommender systems is challenging in complex domains such as travel and tourism. In the case of destination recommendation, no feature set can be readily used as a similarity ground truth, which makes it hard to evaluate the quality of destination characterization approaches. Furthermore, the process should scale well for many items, be cost-efficient, and, most importantly, correct. To evaluate which data sources are most suitable, we investigate 18 characterization methods that fall into the following categories: venue data, textual data, and factual data. We make these data models comparable using rank agreement metrics and reveal which data sources capture similar underlying concepts. To support choosing more suitable data models, we capture the desired concept using an expert survey and evaluate our characterization methods toward it. We find that the textual models to characterize cities perform best overall, with data models based on factual and venue data being less competitive. However, we show that data models with explicit features can be optimized by learning weights for their features.
Linus W. Dietz, Mete Sertkan, Saadi Myftija, Sameera Thimbiri Palage, Julia Neidhardt, and Wolfgang Wörndl. “A Comparative Study of Data-driven Models for Travel Destination Characterization.” In: Frontiers in Big Data 5 (Apr. 2022). ISSN: 2624-909X. DOI: 10.3389/fdata.2022.829939
]]>Tourism is a complex domain for recommender systems because of the high cost of recommending an unsuitable item and the absence of ratings to learn user preferences. Conversational recommender systems have been introduced to provide users with an opportunity to give feedback on items in a turn-based dialog until a final recommendation is accepted. In a scenario such as recommending a city to visit, conversational content-based recommendation may be well-suited since users often struggle to specify their preferences without concrete examples. However, critiquing item features comes with challenges. Users might request item characteristics during recommendation that do not exist in reality, for example, demanding very high item quality for a very low price. To tackle this problem, we present a novel conversational user interface that focuses on revealing the trade-offs of choosing one item over another. The recommendations are driven by a utility function that assesses the user’s preference toward item features while learning the importance of the features to the user. This enables the system to guide the recommendation through the search space faster and accurately over prolonged interaction. We evaluated the system in an online study with 600 participants and found that our proposed paradigm leads to improved perceived accuracy and fewer conversational cycles compared to unit critiquing.
Linus W. Dietz, Sameera Thimbiri Palage, and Wolfgang Wörndl. “Navigation by Revealing Trade-offs for Content-based Recommendations.” In: Information and Communication Technologies in Tourism. Ed. by Jason L. Stienmetz, Berta Ferrer-Rosell, and David Mas- simo. Cham: Springer, Jan. 2022, pp. 149–161. ISBN: 978-3-030-94751-4. DOI: 10.1007/978-3-030-94751-4_14
]]>In this work, we present a data-driven method to mine trips from location-based social networks to understand how tourists travel the world.
The obtained insights can be relevant building blocks for destination recommender systems, i.e., automatic preference elicitation, defining travel regions, and general traveler behavior.
The primary artifact of this paper is the tripmining
library, which quantifies collected trips with several metrics to capture the underlying mobility and assess the quality of the data.
We showcase two applications that utilize the mined trips. The first is an approach for clustering travelers in two case studies, one of Twitter and another of Foursquare, where the pure mobility metrics are enriched with social aspects, i.e., what activities the users have done. Clustering 133,614 trips from Twitter, we obtain three distinct groups of travelers based on the pure mobility trace. In the Foursquare data set, which includes the type of venues the users have checked in, six clusters can be determined. The second application area is the spatial clustering of destinations around the world. These discovered regions are solely formed by the mobility patterns of the trips and are, thus, independent of administrative regions such as countries. We identify 942 regions as destinations that can be directly used as a hierarchical region model in a destination recommender system.
Linus W. Dietz, Avradip Sen, Rinita Roy, and Wolfgang Wörndl. “Mining Trips from Location-based Social Networks for Clustering Travelers and Destinations.” In: Information Technology & Tourism 22.1 (Mar. 2020), pp. 131–166. ISSN: 1098-3058. DOI: 10.1007/s40558-020-00170-6
]]>Linus W. Dietz, Saadi Myftija, and Wolfgang Wörndl. “Designing a Conversational Travel Recommender System Based on Data-driven Destination Characterization.” In: ACM RecSys Workshop on Recommenders in Tourism. Sept. 2019, pp. 17–21
]]>Recommender systems could benefit from not only recommending the most fitting items but also in what quantity the user should consume them. In this paper, we tackle the problem of recommending the personalized duration of stay at a destination. We present a data-driven solution to this problem based on mining trips from location-based social networks. To determine the recommended duration of stay at a destination, we use a statistical approach based on how long travelers typically stay in different cities and how much time the current user generally spends visiting cities. The method can serve as an extension of personalized travel planning systems by not just recommending which city one should travel to but also how much time to spend there.
Linus W. Dietz and Wolfgang Wörndl. “How Long to Stay Where? On the Amount of Item Consumption in Travel Recommendation.” In: ACM RecSys 2019 Late-breaking Results. Sept. 2019, pp. 31–35
]]>In my free time I maintain the bibliography manager JabRef, where we are approaching the JabRef 5.0 major release. Since this will bring substantial changes to the user interface, we want to make sure that we consider the user’s workflow in JabRef. Therefore, we need to learn how the users actually use JabRef!
For this we are supported by Martin Simon, a student of Ergonomics - Human Factors Engineering at the Technical University of Munich, who has developed a short questionnaire on the feature usage of JabRef. This study is conducted within the scope of Martin’s Master’s Thesis in collaboration with the JabRef Development Team. It shall provide insights to the current usage of JabRef that should guide the future development of the user interface. Of course, we will also publish the aggregated results here in the blog.
Follow this link to the anonymous online survey.
If you have any questions or concerns please contact me via email, as I’m a JabRef Maintainer and also advisor to this thesis. For any general JabRef-related issues such as bug reports please use the issue tracker on Github or ask your question in the forums.
]]>We will offer 15 topics, of which I will advise the following:
Some products with recommender systems are used in several contexts, e.g., device type, time of day, etc. The recommendation quality can be improved if the recommender systems is aware of the context and can incorporate this into the ranking of items. It is, however, a challenge to model context effectively and find out the right amount of influence it should have along with traditional features in collaborative filtering and content-based recommendations.
When recommending travel destinations from all around the globe, one needs to have a list of destinations. Often touristic regions correspond to political regions, however this not always the case. This topic is concerned how to find a touristic areas beyond political boundaries and project them on the map.
Researchers in the area of tourist recommendations often face the challenge that tourists do not receive the recommendations at the right time. Although research has shown that context-awareness is important for recommendations on the go, there are several challenges in a tourism scenario. In foreign countries there is often no reliable Internet connection, the user behavior is quite different from their daily life, and sometimes travelers may be uncomfortable to pull out their smartphones, either because they fear to be stolen or it is inappropriate for social reasons. In this topic it is to be explored what would be alternative devices and approaches to receive proactive, context-aware recommendations during travel.
It is said that Facebook, Google and Twitter know you better than yourself. If this was true, it would be a fruitful information source not only for targeted advertisement, but for personalized recommendations in other domains. An exemplary task would how to derive e.g., traveler types from public information of a social media profile. This could include posts, pictures and additional metadata like locations. Neidhardt et al. established the Seven Factor Model for traveler types. Is it possible to reliably derive the traveler type from a social media profile? How could this be implemented in the different social network platforms and evaluated using real persons?
Affective computing is a current hype trend in several disciplines. Analyzing the facial expressions of the user can give clues to her needs, the general context and the physical status, e.g., sleepiness of drivers. Planning a travel is an emotional endeavor. Travel recommender systems could make use of the facial expressions of users to learn what they like. For example, within a small game the user could be presented with some images of travel destinations and by analyzing her reactions using computer vision. This topic would analyze the roots of affective computing and modern solutions to derive emotions from users. Additional focus should be on the design of user interfaces in recommender systems to actually trigger emotional responses by the users while learning about their preferences.
The full information for TUM master students can be found on the course webpage.
]]>I want to use this opportunity to share the ideas behind the book and how everything came together. Some are also described in the acknowledgments and the preface (named “Welcome!”) of our book, which are available as a preview at the publisher’s homepage.
So, why is there a need for another Java book? After all, there are established classics on code quality like Clean Code or Refactoring, and recently Joshua Bloch released the third edition of Effective Java. When telling our idea to one of my professors in Bamberg, he told me “…that he wouldn’t waste a single minute on writing a Java book, because there are so many excellent books out there and writing another one would be rockstar business”. That was half a year before we signed with PragProg, our first choice publisher.
I think Java by Comparison is different, because of its didactic concepts that make it accessible for learners that are not yet on a professional programming level. Although we love the aforementioned books and have learned so much from them, they are not quite an easy read for a programming novice. Our book basically condenses our combined experience in teaching advanced Java programming courses at university. Having mastered the basic syntax is just not enough to write good code. This needs a professional environment, where code is not only written once and forgotten after the assignment submission (as it often happens at university courses), but is maintained in a team over a prolonged period. It also needs somebody with experience to review the code and give you hints on how to improve it.
In an ideal case, you have a mentor! Somebody who points out and explains to you your mistakes, shows you how to resolve them and gives you an idea why the improvement is actually better than what has been there before. Unfortunately, good mentors are rare and usually have little time. So we wrote a book, which we think can help to learn about Java code quality in the absence of a mentor.
This tweet by David Heinemeier Hansson (the creator of Rails) inspired us to structure our items in the two-page format. When you open the book you will find some flawed code on the left side along with an explanation which is then contrasted to the improved code together with an rationale why the solution is better. On top of this there’s an actionable name (like “Avoid Negations”) that sticks in mind.
Often novices are told to improve their programming skills by reading code written by experts. While in theory it is possible to go to Github and start reading code of open source software, in practice a 2nd year programmer can’t do that without guidance. Furthermore, much code out there is actually of lacking quality. But how can a novice know the difference..?
With the 70 two-page comparisons in nine chapters you will read some code, but not randomly. Instead, each snippet focuses on only one issue that you will learn to memorize and generalize to other code. Plus, you have the direct comparison on how to improve the flawed code, meaning that you not only learn what is bad and good code, but also how to get from the former to the latter.
Thus, this book is also good for junior developers that are about to start their professional programming careers. Before publishing it, we sent it to about 20 technical reviewers both from academia and software industry who thoroughly went through the chapters and reported many issues that were mistakes or where they simply disagreed with our opinion. While their comments improved the quality of the book a lot, they also motivated us to keep going, because several said that they really needed such a book for their students, junior developers and even project contractors.
Enough said: You can buy the book and the DRM-free e-book directly at Pragmatic Programmers, or if you prefer Amazon.com [UK, DE], also over there. We also created a small webpage with more condensed information at java.by-comparison.com.
]]>