Individuals on the move, e.g., tourists on a sightseeing trip in an unfamiliar city often find themselves overwhelmed by the challenges of coping with unfamiliar environments. This presents a need for tools and methods that will guide them by providing them useful recommendations while they are "on the move." Recent advances in mobile and sensor-based technologies have made it possible to collect and process location traces across many different mobile applications. Such data, when combined with other spatio-temporal, contextual, and user-specific information can, in principle, be used to generate useful recommendations for individuals on the move.
This exploratory research project formulates and explores a novel variant of recommender systems, namely, mobile sequential recommender systems for mobile users where each recommendation takes into account the trajectory and history of past recommendations, as one of selecting a sequence of locations to recommend under a set of spatio-temporal, contextual, and privacy constraints. Given the combinatorial nature of the problem (where the size of the search space grows is exponential in the relevant parameters) the project aims to explore heuristics. It will also develop appropriate measures for assessing the effectiveness of alternative solutions.
The project, if successful, would establish the feasibility of a line of investigation that could lead to the development of effective approaches to sequential recommendation problem with obvious benefits to mobile users. The project enriches research based advanced training opportunities for graduate and undergraduate students. All of the data, software, and publications resulting from the project will be made freely available to the broader research community.
The major focus of this project is on studying the unique characteristics of mobile recommender systems and on developing approaches to providing personalized mobile recommendations to the users. In particular, this project focused on the following two problems: how to (a) incorporate costs into the recommendation process in the context of recommending travel tours to the customers, and (b) provide real-time recommendations of driving directions to the taxi, limousine and ride-sharing drivers (when they don’t have passengers) to maximize their chances of picking customers in their current locations and circumstances. To address the first problem of recommending travel tour packages, we have first defined and modeled user cost preferences and incorporated them into three different cost-aware latent factor recommendation models. Then we validated and compared these three models on a real-world travel tour data and showed that all the proposed cost-aware recommendation models consistently outperformed existing latent factor models (that do not explicitly model costs) by a significant margin. Furthermore, we have compared the three proposed models among themselves and showed that the two particular models (i.e., the extended Maximum Margin Matrix Factorization and the Logistic Probabilistic Matrix Factorization models) lead to better performance improvements than the third model. To address the second problem of providing real-time recommendations to the taxi, limousine and ride-sharing drivers on where they should go in the city to have a better chance of picking up clients so that they can maximize their expected earnings, we have developed an algorithm that provides such recommendations. The proposed method utilizes the "heatmap" approach in which a map of the city is presented to the drivers with various regions being colored differently according to the propensity of picking up clients in each region. We have also tested the proposed method on real drivers of a ride sharing company by conducting a randomized controlled experiment (the so called, A/B testing). Unfortunately, our algorithm did not lead to statistically significant performance improvements. Nevertheless, we have learned a lot about reactions of the drivers to our recommendations and the human factors and trust-related aspects of the recommendation process in this project. All this provided valuable additional insights into possible future directions of our research in this area, especially pertaining to the human factors aspects of recommendations. While working on this project, a graduate student has received extensive training in the fields of recommender systems, machine learning and big data. As a result of all this training and work on the project, the student was hired by a major ridesharing company to work on the problems that are directly related to the ones studied in this project. Furthermore, the company became interested in some of the ideas developed in the project.
