Project: Mobile User Modeling

People: Dr. Wolfgang Wörndl, Hubert Kreuzpointner, Prof. Dr. Johann Schlichter

Smartphones, PDAs (personal digital assisants) and other mobile devices are becoming more and more powerful and are increasingly used for tasks such as searching and browsing Web pages, or managing personal information.  However, mobile information access still suffers from limited resources regarding input capabilities, displays, network bandwidth etc. Therefore it is desirable to assist the user by services that are tailored towards her context such as her current location. In this project, we investigate selected research problems regarding mobile user modeling and personalization, for example recommender systems in mobile scenarios. In addition, it is important to consider privacy concerns of users, since (location-based) services use the current user location and other sensitive information to personalize information access.

Related project: Context-aware Recommender Systems in Mobile Scenarios

List of currently available thesis topics by our chair.

Diplomarbeit, Bachelor or Master Thesis in this area.

Subproject: Mobile User Activity Logging (MUAL)

People: Dr. Wolfgang Wörndl, Florian Schulze

Three main steps can be identified in a general user modeling process: 1. Collecting data about the user, 2. Analyzing the data to build a user model, 3. Using the user model to adapt information access. In this subproject, we focus on the first step of this user modeling process, the collection of data about the user in a mobile environment. Existing systems only handle specific information or sensors only (e.g. analysis of user location based on GPS logs), or were created for different purposes. The goal of our work is a unified approach for collecting data about user actions on mobile devices in an appropriate granularity for user modeling.

We have designed and implemented a framework for mobile user activity logging on Windows Mobile PDAs. The framework is based on the MyExperience project as a foundation for our implementation. MyExperience aims at collecting data for allowing early detection of subject compliance or technology issues of mobile software. We have extended this system with hardware and software sensors to monitor phone calls, messaging, peripheral devices (e.g. BT headsets), media players, GPS sensor, networking, personal information management (e.g. contacts and appointments), web surfing, system behavior (e.g. power status) and usage of arbitrary applications. It is possible to detect when, at which location and how a user uses an application or accesses certain information, for example. For evaluation we tried our framework in several usage scenarios and were able to validate that our framework was able to collect meaningful information about the user without obstructing her too much. Next steps include analyzing the collected data and building the user model, and thus carrying out the second step in the user modeling process.

Subproject: Mobile Episodic Information Management

People: Dr. Wolfgang Wörndl, Dr. Georg Groh

Smartphones are getting increasingly powerful and become more and more part of people’s
everyday life. As most people carry their smartphones with them at all times, they can record
noteworthy moments right as they happen. We call the record of such an event an episode.
The goal of this subproject is to examine how episodic information management on common
smartphones can be supported to help people building a personal memory.

We have designed and implemented a system to record episodes on Android smartphones.
Our approach allows for capturing all relevant information of an episode while keeping manual
interaction as simple and straight-forward as possible. Episodes are enhanced with context
information such as the current location or a picture, for example. The application was evaluated
in a small user study with promising results. Current work focuses on advanced options for
searching, filtering and visualizing existing episodes.

Subproject: Location Privacy

People: Hubert Kreuzpointner

LocPri is an approach to ensure privacy for users of mobile location based services. An increasing number of mobile phone and PDA users has access to a widespread communication infrastructure enabling them to use internet services whenever and wherever they want. The software industry pursues the development of such services with great effort because a large economic potential has been forecast for them. By using LBS, it is possible to collect movement profiles of users and to derive other high-level information that can be used to compromise the privacy of the user.  

(Concluded) Subproject: SeMoDesk - Mobile Semantic Personal Information Management

People: Dr. Wolfgang Wörndl, Florian Schulze

SeMoDesk is an approach to support personal information management on PDAs based on Semantic Desktop ideas. Users can define and manage their personal ontology (left screenshot), and assign relations between resources. To do so, SeMoDesk assists the user as much as possible, for example calls or short messages on the mobile device are integrated automatically. It is possible to structure and browse the personal information space across different applications. We have realized a recommendation function for finding resources that may be of current interests to users (middle screenshot). In addition, SeMoDesk offers a feature to show information about relevant point-of-interests on a map, together with the location of upcoming appointments (right screenshot). Other work is concerned the integration with an RFID infrastructure to improve the context-sensitivity of the system, and learning relevant user locations based on GPS logs.

Selected Publications

 Capturing, Analyzing and Utilizing Context-Based Information About User Activities on Smartphones

 Modeling and Learning Relevant Locations for a Mobile Semantic Desktop Application

 A Framework for Mobile User Activity Logging

 Issues in Mobile User Modeling and Search Personalization

 SeMoDesk: Towards a Mobile Semantic Desktop

 Generating Virtual Users with Real Path Information to Improve Location Privacy

All publications of our chair