Master Thesis: Analyzing Social Network and Smartphone Data to Model Contextual Contact Interactions
Analyzing Social Network and Smartphone Data to Model Contextual Contact Interactions
Social networks, such as Facebook or Google+, enable users to model their social relations to other members of these networks. Unfortunately, these relations are often inaccurate and incomplete as they do not reflect how closely two individuals are actually related. Facebook, for example, tries to solve this problem with grouping features, where users are added to user-defined groups. Google+ introduces the concept of circles in order to categorize people. However, these methods require manual maintenance effort and do not generally describe the distance of relationships between friends and groups.
There is a lot of contextual information present in social networks as well as mobile phones that can help to describe the contextual contact interactions of users to other users or groups. Mobile phone data, such as the call history, messages and emails sent correlated to social network information, such as the number of interactions between users, friend categories or the activity in different social networks, can help to build dynamic and spontaneous social networks, which can provide an added value to other services, such as recommender systems.
The main objective of this thesis is to analyze and extract relevant information from smartphones as well as different social networks in order to describe, model and categorize contextual contact interactions between users. These models can then be evaluated and compared to circles of friends in reality by taking a closed group of users into consideration.
The basic objectives of this thesis are:
- Develop a service that analyzes a user's communication history (phone calls, SMS/IM/email logs, and else) running on an Android-based smartphone.
- Enrich the created model with context information acquired from social networks.
- Aggregate this information in order to model the dynamic and spontaneous social network for a given user.
- Analyze implicit information about the relations of the users in the modeled meta-network (e.g., is someone a colleague or a friend).
- Evaluate and compare the results to circles of friends in reality.
- Apply data mining techniques to analyze the communication protocols
- Integrate models of social groups from multiple sources
- Detect "false friends", i.e., people who are not really related in any way to an individual
- Deal with privacy related issues
- Knowledge of Web APIs
- Knowledge of mobile telephony technologies and mobile operating systems might be useful
- Interest to work in the field of mobile computing, social network analysis and data mining
Supervisor: Prof. Dr. Axel Küpper , Sebastian Göndör 
Type: Master Thesis
Duration: 6 months
10587 Berlin, Germany
Phone: +49 30 8353 58811
Fax: +49 30 8353 58409