MT/BT Exploring SOLID as a Platform for Decentralized Personal Recommendations

Recommender systems recommend movies, restaurants or other items to an active user based on information about users and items. Usually, recommender systems are centralized, i.e. user ratings and all other information is stored and managed on centralized servers. However, the idea of a decentralized recommender system has some advantages especially in mobile scenarios. In this case, users would manage their preferences, ratings and other personal information on their mobile devices to improve user control and reduce privacy concerns. The information can then be selectively shared with other users and server-based applications to generate personal recommendations locally on the user device.

Recently, SOLID has been proposed as a specification to store personal information in decentralized data stores called Pods. Implementations allow to control access to these Pods and implement a vision of a decentralized Web (https://solidproject.org/). SOLID could a suitable framework for decentralized recommender systems as well. The scenario is that users manage their preferences in a local Pod (e.g. about preferred restaurants) on their mobile device and generate recommendations about interesting items in the vicinity (e.g. nearby recommended restaurants), without revealing their preferences to centralized services. Linked open data has also been proposed in combination with decentralized data storage, for example to build and exploit a (Personal) Knowledge Graph. So the goal of the thesis project is to explore SOLID as a platform for decentralized, possibly semantic, recommender systems.

The proposed course of action is as follows:
* Overview of the state-of-the-art in (decentralized) recommender systems and decentralized data storage related to SOLID
* Designing a solution and implementing a prototype for the explained scenario
* Evaluating the solution, e.g. in an offline study and/or from a user's perspective

Prerequisites are high motivation and good enough programming skills for the prototype. Please send your application (brief CV, transcript of records and short motivation statement) to Wolfgang Wörndl (woerndl@in.tum.de) until Oct. 3rd.