EAGER: programming repetitive tasks by example
CHI '91 Proceedings of the SIGCHI Conference on Human Factors in Computing Systems
Analysis of user behaviour as time series
HCI'92 Proceedings of the conference on People and computers VII
Pattern-matching and text-compression algorithms
ACM Computing Surveys (CSUR)
Efficient string matching: an aid to bibliographic search
Communications of the ACM
Programming Applications for Microsoft Windows with Cdrom
Programming Applications for Microsoft Windows with Cdrom
Gesture Modeling and Recognition Using Finite State Machines
FG '00 Proceedings of the Fourth IEEE International Conference on Automatic Face and Gesture Recognition 2000
GUI Ripping: Reverse Engineering of Graphical User Interfaces for Testing
WCRE '03 Proceedings of the 10th Working Conference on Reverse Engineering
Detours: binary interception of Win32 functions
WINSYM'99 Proceedings of the 3rd conference on USENIX Windows NT Symposium - Volume 3
Empirical analysis of predictive algorithms for collaborative filtering
UAI'98 Proceedings of the Fourteenth conference on Uncertainty in artificial intelligence
Journal of Network and Computer Applications
A probabilistic risk analysis for multimodal entry control
Expert Systems with Applications: An International Journal
Hi-index | 12.05 |
This paper describes a means of unsupervised learning of recurring patterns in user activity through patterns in system level events generated by a graphical user interface. Earlier work has shown that using this distillation of the more complex behavioural interaction between the user and the application provides a symbolic representation of knowledge and goals that could be used to imply preference. Although prior research has explored the possibilities of removing this information acquisition bottleneck in such an expert system using ambient monitoring approaches, some have experienced difficulty in dealing with the varying length training sequences and segmentation of the continuous event stream. Unlike previous work the approach documented here handles interactions of varying sizes and is able to recall recurrent patterns in real time irrespective of the number of interactions learned. In addition to describing the proposed approach we also describe the shortcomings of various previously applied machine learning techniques on the same type of data. We also demonstrate a practical implementation of our approach applied to web browser usage.