Inhalt des Dokuments
Es gibt keine deutsche Übersetzung dieser Webseite.
Bachelor Thesis: TYDR-Fusion: Ubiquitous Social Networking based on Automatically Tracked Data
TYDR-Fusion: Ubiquitous Social Networking based on Automatically Tracked Data
Due to the more and more powerful hardware, modern smartphones are equipped with every year, the idea of ubiquitous computing became a reality. In combination with data provided by location-based social networks and the technology behind recommender systems, the concept of ubiquitous social networking can be enabled.
Ubiquitous social networking presents a new way of group interactions, a way where participants do not need an Internet connection in order to share their information. They can do it almost everywhere and on devices that most people have access to. Moreover, social interactions are enhanced by recommender systems that provide users with recommendations based on their group’s preferences.
In this thesis, I present a solution for enabling this new concept by designing and implementing two group modules. Each module allows multiple users to share data in a device-to-device (D2D) way and provides group-specific recommendations based on the shared data. The first module focuses on music data. Users have the opportunity to share their music preferences in order to receive a playlist that should fit the whole group’s taste. The second module focuses on location data. Each group member can share their locations history in order to receive a list of recommended restaurants that are close to all members. Both modules specialize in different scenarios but have the same goal, to enhance networking within a group of people.
The foundation of my solution is device-to-device data exchange. Such an approach assures high ubiquity and privacy. Based on research presented in this thesis, I have decided to use Google Nearby Connections as D2D library. Unfortunately, this library comes with some issues, which are pointed out in my work.
Recommender systems play a significant role in my solution. I have researched such systems in the context of point of interest (POI) and group music recommendations. Each recommender supports different input data structure. In order to generate recommendations based on information shared by group members, I have designed and implemented custom processing algorithms. The performance of these algorithms have been tested, and the results have been evaluated.
According to my test results, three people need on average a little less than 20 seconds in order to exchange their music preferences (100 tracks each) and receive a recommended playlist. Using the group locations module on the other hand to receive a list of recommended restaurants based on the members’ most visited locations takes only a little bit more than 3 seconds.
Supervisor: Felix Beierle
Type: Bachelor Thesis
Duration: 4 months
10587 Berlin, Germany
Phone: +49 30 8353 58811
Fax: +49 30 8353 58409