Computing Reviews
Today's Issue Hot Topics Search Browse Recommended My Account Log In
Review Help
Search
User-centric inference based on history of context data in pervasive environments
Kalatzis N., Roussaki I., Liampotis N., Strimpakou M., Pils C.  Services integration in pervasive environments (Proceedings of the 3rd International Workshop on Services Integration in Pervasive Environments, Sorrento, Italy, Jul 7, 2008)25-30.2008.Type:Proceedings
Date Reviewed: Sep 1 2008

Intelligent pervasive computing systems (IPCS) are useful for forecasting the situational information of users, such as their daily routine, and activities at any time or location. The literature discusses the architecture of an IPCS equipped with situation-aware agents for predicting and alerting the flow of debris disasters [1]. Yet, the design and implementation of proactive IPCS for user-centered information inference by means of historical context data (HCD) pose difficult questions. How should an IPCS dynamically detect and adapt to the environments, resources, and preferences of users? How should the principal context parameter be targeted for defining the prediction rules of context estimation?

Kalatzis et al. propose a framework for storing and modeling the HCD used to forecast the context and location information of users. The framework includes an algorithm for periodically caching context information into an HCD database (HCDDB), and procedures for creating and maintaining the HCDDB based on adjustable threshold time frames and the relevance of prediction rules to the activities and locations of users. The incidences of each context data combination in a time frame are assigned a daily summative attenuation score, and stored as a single entry in the HCDDB. The maximum score, the minimum threshold score, the minimum score difference of entries, and the cumulative scores of selected entries in a time frame are alternative models available in context estimation. The probabilities of a user performing specific activities in time frames are computed from the entries in the HCDDB. These probabilities are used in context prediction. The prediction rules are responsive to the constraints of the resources and the requirements of the context estimation.

An experiment was performed to evaluate the effectiveness of the HCD framework in predicting user-centered information. A random sample of volunteers carried cell phones equipped with software designed to identify their movement locations over 13 weeks. The initial 12 weeks yielded the training dataset for a context inference engine. The experimental context information was collected in the last week. The evaluation results revealed significant user-location successes, and an excellent storage reduction in the volume of data in the HCDDB. Consequently, the advocated methodology for user-centered information deduction with HCD is useful for predicting subjective types of context information.

Reviewer:  Amos Olagunju Review #: CR136010 (0910-0974)
1) Kung, H.; Ku, H.; Wu, C.; Lin, C. Intelligent and situation-aware pervasive system to support debris-flow disaster prediction and alerting in Taiwan. Journal of Network and Computer Applications 31, 1(2008), 1–18.
Bookmark and Share
  Reviewer Selected
Featured Reviewer
 
 
Inference Engines (I.2.3 ... )
 
 
Data Compaction And Compression (E.4 ... )
 
 
Learning (I.2.6 )
 
 
Models (I.5.1 )
 
 
Software Architectures (D.2.11 )
 
 
Systems And Software (H.3.4 )
 
  more  
Would you recommend this review?
yes
no

E-Mail This Printer-Friendly
Send Your Comments
Contact Us
Reproduction in whole or in part without permission is prohibited.   Copyright 1999-2024 ThinkLoud®
Terms of Use
| Privacy Policy