| Project Title: | Handling Concept Drift in Adaptive Information Systems |
| Duration: | 2008-2012 |
Summary
One of the main goals in Adaptive Information Systems (AIS) is to serve the information need of users by adaptively filtering content based on the users' interests, knowledge, background, goals and tasks. User Modeling (UM) is an active research area that studies how a machine can learn, model and make use of a representation of the users' state of mind (and other factors). An AIS only has partial information about the users as it can only observe their behavior. The AIS is thus performing UM under the presence of (many) hidden contexts.
In this project we deal specifically with the problem of detecting changes in the goals or interests of users, which corresponds to the problem of concept drift (CD) in Data Mining (DM) and Machine Learning (ML). Modern DM/ML techniques that address CD have so far had only modest success in real AIS applications. We believe that this is mainly due to the fact that the state-of-the-art ML approaches do not fully take into account several factors which the UM community is aware of: the development of the scope of user-interests, the hierarchical nature and rich semantics of information content, and existence and evolution of the user social networks.
This project is aimed at development of a unifying framework and corresponding DM techniques for handling CD in AIS, evaluation of the framework and techniques, first through various simulation studies, and through integration of the techniques with existing AIS for the external validation of our work.
Research Team
|
Visitors and Collaboration
- Dr. Indrė Žliobaitė
- Dr. Albert Bifet, Waikato University, New Zealand, August 2010;
- Dr. Oleksiy Mazhelis, University of Jyvaskyla, Finland, August 2010;
- Dr. Carlos Soares, University of Porto, Portugal, November 2009.
- CFB Boilers Sensors Data Mining, VTT, Finland;
- Sligro B.V., Food Sales Prediction.
|
Events
- Handling Concept Drift in Medical Applications: Importance, Challenges and Solution, Tutorial at IEEE ACM Symposium on Computer Based Medical Systems (CBMS 2010), Perth, Australia, October 2010;
- Handling Concept Drift in AIS: Importance, Challenges and Solutions (HaCDAIS 2010)Workshop at ECML/PKDD'2010, Barcelona, Spain, September 2010;
- DH Group workshop on Detecting and Handling Drift (DHDHD 2010), Eindhoven, August 2010.
Publications
- Žliobaitė, I., Bakker, J., Pechenizkiy, M. (2009). OMFP: An Approach for Online Mass Flow Prediction in CFB Boilers. Discovery Science 2009: 272-286. [PDF] [BIB]
- Andriy Ivannikov, Mykola Pechenizkiy, Jorn Bakker, Timo Leino, Mikko Jegoroff, Tommi Kärkkäinen, Sami Äyrämö. 2009. Online Mass Flow Prediction in CFB Boilers. ICDM 2009: 206-219. [PDF] [BIB]
- Jorn Bakker, Mykola Pechenizkiy, Indre Zliobaite, Andriy Ivannikov, Tommi Kärkkäinen. 2009. Handling outliers and concept drift in online mass flow prediction in CFB boilers. KDD Workshop on Knowledge Discovery from Sensor Data 2009: 13-22 (Best paper award)
[PDF] [BIB]
-
Žliobaitė, I., Bakker, J., Pechenizkiy, M. (2009). Towards context aware food sales prediction. In Y. Saygin et al. (Eds.), Proceedings 3nd International Workshop on Domain Driven Data Mining (DDDM'09), in: IEEE International Conference on Data Mining: Workshops (ICDM'09, Miami, Florida, USA, December 6-9, 2009). (pp. 94-99). [PDF] [BIB]
-
Bakker, J., Pechenizkiy, M. (2009). Food wholesales prediction : what is your baseline? In J. Rauch, Z.W. Ras, P. Berka, T. Elomaa (Eds.), Foundations of Intelligent Systems (18th International Symposium, ISMIS 2009, Prague, Czech Republic, September 14-17, 2009. Proceedings). (Lecture Notes in Computer Science, Vol. 5722, pp. 493-502). Berlin: Springer. [PDF] [BIB]
-
Meulstee, P., Pechenizkiy, M. (2008). Food sales prediction: "If only it knew what we know". Proceedings 2nd International Workshop on Domain Driven Data Mining (DDDM'08), in: IEEE International Conference on Data Mining: Workshops (ICDM'08, Pisa, Italy, December 15-19, 2008). (pp. 134-143). IEEE Computer Society. [PDF] [BIB]
Unpublished Technical Reports
-
Žliobaitė, I., Bakker, J., Pechenizkiy, M. (2010). Beating the baseline prediction in food sales: How intelligent an intelligent predictor is? (SIGKDD2010 Submission, Industry track).
- Žliobaitė, I., Bakker, J., Pechenizkiy, M. (2010). Context Aware Sales Prediction: experimental evaluation. (TR-
Feb2010.pdf).
-
Žliobaitė, I., Bakker, J., Pechenizkiy, M. (2009). Towards Context Aware Sales Prediction. (TR-Sep2009.pdf, exte
nded version of DDDM'09 submission).
-
Žliobaitė, I. (2009). On Recognition of Seasonal Predictability in Food Product Sales. (TR-Jul2009.pdf).
Presentations and posters
|