Hybrid Solutions to the Feature Interaction Problem

Muffy CALDER, Mario KOLBERG, Evan MAGILL, Dave MARPLES, and Stephan REIFF-MARGANIEC

University of Glasgow
Computing Science
17 Lilybank Gardens
Glasgow G12 8RZ, UK
muffy@dcs.gla.ac.uk
 University of Stirling
Computing Science & Mathematics
Stirling FK9 4LA, UK
{mko, ehm, srm}@cs.stir.ac.uk
 Global Inventures Inc
Bishop Ranch 2
2694 Bishop Drive, Suite 27
San Ramon, CA, 94583
dmarples@iee.org

Abstract. In this paper we assume a competitive marketplace where the features are developed by different enterprises, which cannot or will not exchange information. We present a classification of feature interaction in this setting and introduce an online technique which serves as a basis for the two novel hybrid approaches presented. These approaches are hybrid as they are neither strictly off-line nor on-line, but combine aspects of both. The two approaches address different kinds of interactions, and together they provide a complete solution by addressing interaction detection and resolution. We illustrate the techniques within the communication networks domain.