SDA SE Wiki

Software Engineering for Smart Data Analytics & Smart Data Analytics for Software Engineering

User Tools

Site Tools


Differences

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

Link to this comparison view

private:daniel:public:scam2018:start [2018/06/11 15:52] (current)
Daniel Speicher created
Line 1: Line 1:
 +====== SCAM 2018: Code Smell Detection with Reason ======
 +
 +===== Abstract =====
 +Automatic ​ code  smell  detection ​ could  be  of  valuable ​ help  to
 +developers if it  would not raise false alarms. ​  Every single false
 +positive diminishes the trust a developer has in the smell detection
 +and increases her habituation to unjustified warnings.
 +
 +To  deepen our  understanding ​ of the  phenomenon ​ "bad smell  false
 +positive", ​ we designed ​ a study  that  deliberately led  to a  high
 +number of false positives: We applied a rather rigid classical smell
 +definition to  thoughtfully designed software. ​ Each  potential flaw
 +was then systematically and carefully evaluated to decide whether it
 +should ​ be refactored ​ or not.   For false  positives we  identified
 +relevant developer concepts that were  not accounted for by the code
 +smell.
 +
 +We found that developers had traded the smell for design concepts of
 +higher ​ value. Half  of these  concepts were  constructive concepts,
 +i.e. they  can not be  sufficiently captured by  analytical criteria
 +and have  thus been missed ​ by previously suggested ​ smell detection
 +approaches. ​ Surprisingly ​ the original smell definition ​ turned out
 +to be even incomplete without the incorporation of some constructive
 +concepts.
 +
 +Although the constructive concepts can not be captured in analytical
 +terms, ​ human developers ​ recognize ​ them based  on identifiers ​ and
 +common ​ developer knowledge. ​  To the  degree that  these identifier
 +conventions and developer knowledge is generally accepted, it can be
 +encoded into  the detection rule.   To the degree that  the concepts
 +are specific to a context, the detection rules can be adapted within
 +the context.
 +
 +Constructive ​ concepts play  a pivotal ​ rule in  the definition ​ and
 +adaptation of  bad code  smells. Encoding ​ them explicitly ​ in smell
 +definitions ​  ​captures ​ developer'​s ​  ​reasoning ​ and   ​substantially
 +increases the precision.
 +
 +
 +
 +===== Links =====
 +  * http://​www.ieee-scam.org/​2018/​
 +
 +===== Accompanying data =====
 +  * TBD
 +
 +
  
private/daniel/public/scam2018/start.txt · Last modified: 2018/06/11 15:52 by Daniel Speicher

SEWiki, © 2018