Version 4 (modified by acher, 7 years ago)


Merge Operators for Feature Models: Experimental Results

The following web page provides material, details and results of the experiments we conduct in merging feature models. The web page is related to a paper submitted to  JOT journal, entitled "Managing Feature Models Using Merging Techniques", co-authored by  Mathieu Acher, Philippe Collet],  Philippe Lahire and  Robert B. France.

Input feature models to be merged

Examples of feature models (FMs) randomly generated are given below:

nFM being the number of feature models to be merged, nComm the number of features commonly shared by the input FMs and per the percentage of features commonly shared by the input FMs (e.g., for nFM = 10, nComm = 100, per = 80, 20*10=200 features are not commonly shared by the input FMs). We assume that two features are similar/shared if their names are strictly equal.

Other feature models for nFM 10..250 are available on demand (the entire archive is more than 20 Go!).

Experimental results

The zip archive is available  here

There are two folders in the archive:

  • commonProducts: results for the merge in intersection mode (note that the number of nodes may be equal to be zero in case there is no common configurations between FMs).
  • competing: results for the merge in strict union mode

Three files are available for the different values of nFM (10 <= nFM <= 250):

  • nSPLx.pdf: calculation time vs number of features
  • nSPLx-sc.pdf: number of nodes vs calculation time
  • nSPLx-space.pdf: number of nodes