Software Clustering

User Tools

Site Tools

You are here: Welcome to the Software Clustering Wiki » LimSim
Trace:
refsim

Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Both sides previous revisionPrevious revision
Next revision		Previous revision
refsim [2010/03/21 20:26] – markrefsim [2010/05/06 13:53] (current) – bil
Line 1: Line 1:
 ====== LimSim ====== ====== LimSim ======
 +
 ====== Overview ====== ====== Overview ======
-The LimSim allows to evaluate software clustering algorithms on large number of simulated factbases. It creates a large number of simulated software systems with available authoritative decompositions and then evaluates software clustering algorithms on those systems using an existing evaluation distance. 
  
-The input parameters for LimSim method are  MDG, an authoritative decomposition and evaluation distance. The algorithm applies a series of small modifications to the MDG with corresponding modifications for the authoritative decomposition. Each modification is designed in such a way as to ensure that the resulting decomposition is indeed authoritative for the resulting MDG.+LimSim allows the evaluation of software clustering algorithms on a large number of simulated factbases. It creates a large number of simulated software systems with available authoritative decompositions and then evaluates software clustering algorithms on those systems using an existing evaluation distance. 
 + 
 +The input parameters for LimSim are an MDG, an authoritative decomposition and an evaluation measure. The algorithm applies a series of small modifications to the MDG with corresponding modifications for the authoritative decomposition. Each modification is designed in such a way as to ensure that the resulting decomposition is indeed authoritative for the resulting MDG.
  
 After a large number of randomly selected modifications has been applied, we will have an MDG that is significantly different than the original one for which an authoritative decomposition exists. By repeating this randomized process many times, we can obtain a large After a large number of randomly selected modifications has been applied, we will have an MDG that is significantly different than the original one for which an authoritative decomposition exists. By repeating this randomized process many times, we can obtain a large
 population of simulated software systems with available authoritative decompositions. population of simulated software systems with available authoritative decompositions.
  
-Finally, LimSim calculates quality of a software clustering algorithm on those generated MDGs.+Finally, LimSim calculates the quality of the software clustering algorithm on the generated MDGs by using a measure, such as MoJoFM. 
 + 
 +The current LimSim implementation includes the following 5 modifications:
  
-LimSim implementation includes 5 such modifications presented below. 
   - Merge two modules   - Merge two modules
   - Split a module   - Split a module
Line 24: Line 27:
   -  Run application using the following command   -  Run application using the following command
  
 +       java limsim.LimSim  
 +
 +====== Operation Instructions ======
 +
 +The LimSim application has the following command line parameters:
  
-       java limsim.LimSim  -auth <decomposition> -mdg <MDG> -eval <evaluation distance> algorithm  algorithm algorithm [-loadConfig file] [-saveConfig File] [-saveConfigXsd configSchema] [-log logConfigFile] [-iter number]+   java limsim.LimSim  -auth <decomposition> -mdg <MDG> -eval <evaluation distance> algorithm  algorithm algorithm [-loadConfig file] [-saveConfig File] [-saveConfigXsd configSchema] [-log logConfigFile] [-iter number]
  
-====== Operation Instruction ====== 
 ===== Input parameters ===== ===== Input parameters =====
  
   * -auth an authoritative decomposition stored in rsf format   * -auth an authoritative decomposition stored in rsf format
   * -mdg a module dependency graph stored in rsf format   * -mdg a module dependency graph stored in rsf format
-  * -eval  class name which implements an evaluation distance. Currently, it is implemented MoJoFM (jret.evaluation.decomposition.MoJoFMCalculator) and KE(jret.evaluation.decomposition.KoschkeEvaluator) +  * -eval  class name which implements an evaluation distance. Currently supported measures are MoJoFM (jret.evaluation.decomposition.MoJoFMCalculator) and KE(jret.evaluation.decomposition.KoschkeEvaluator) 
-  * -loadConfig name of configuration file. The configuration file contains JRET parameters such number entities, list of modification operations etc.+  * -loadConfig name of configuration file. The configuration file contains JRET parameters such number of entities, list of modification operations etc.
   * -saveConfig saves JRET configuration parameters.   * -saveConfig saves JRET configuration parameters.
   * -saveConfigXsd saves schema of configuration parameters   * -saveConfigXsd saves schema of configuration parameters
Line 41: Line 48:
 ===== Output ===== ===== Output =====
  
-LimSim prints evaluation information for each software clustering algorithm. This is included: +LimSim prints evaluation information for each software clustering algorithm. This includes: 
-- Average evaluation value +  * - Average evaluation value 
-- Standard deviation of evaluation values +  * - Standard deviation of evaluation values 
-- The worst evaluation result +  * - The worst evaluation result 
-- The best evaluation result+  * - The best evaluation result
  
refsim.1269203179.txt.gz · Last modified: 2010/03/21 20:26 by mark