Beta 1


Title Context Mining
Author Ciosek, Tomasz Lukasz
Supervisor Larsen, Jakob Eg (Cognitive Systems, Department of Informatics and Mathematical Modeling, Technical University of Denmark, DTU, DK-2800 Kgs. Lyngby, Denmark)
Petersen, Michael Kai (Cognitive Systems, Department of Informatics and Mathematical Modeling, Technical University of Denmark, DTU, DK-2800 Kgs. Lyngby, Denmark)
Institution Technical University of Denmark, DTU, DK-2800 Kgs. Lyngby, Denmark
Thesis level Master's thesis
Year 2010
Abstract This thesis aims to answer the question if and how the context of a mobile user can be identified based on raw data gathered with mobile phone’s sensors. The aim is to recreate discrete contexts without any user input, meaning that all methods used should unsupervised. The context in question is general information, like location, movement, surrounding devices; describing the environment of the user, based on built‐in sensors of a typical mobile phone. The continuous timeline of sensor readings is split into various groups (can be many at the same time), which indicate that at the given point of time a certain set of patterns associated with the groups has been identified. In order to increase the feasibility of the study, a high focus was put on mining location data, as it is seen as most important indication of context after time dimension. Additionally taking into account battery requirements, particular focus was put on mining GSM data, as this sensor is the only one that is turned on all time in every mobile phone.
Imprint Technical University of Denmark (DTU) : Kgs. Lyngby, Denmark
Series IMM-M.Sc.-2010-40
Fulltext
Original PDF ep10_40_small.pdf (14.51 MB)
Admin Creation date: 2010-06-08    Update date: 2010-06-08    Source: dtu    ID: 263600    Original MXD