Assessing Single-Objective Performance Assessing Single-Objective - - PowerPoint PPT Presentation
Assessing Single-Objective Performance Assessing Single-Objective Performance Convergence and Time Complexity for Convergence and Time Complexity for Refactoring Detection Refactoring Detection by D. Nader-Palacio, D. Rodriguez -Cardenas , J.
by
Universidad Nacional de Colombia & Universidad del Rosario Research Group on Artificial Life (Alife) GECCO 2018 Kyoto
[Software Refactoring] consists of re-constructing the code design of a [Software Refactoring] consists of re-constructing the code design of a software system without affecting the software system without affecting the behavior functionality behavior functionality (Fowler & (Fowler & Beck, 1999) Beck, 1999)
[Refactoring & Reconstruction] Refactoring is a subset of [Refactoring & Reconstruction] Refactoring is a subset of Reconstructions Reconstructions (Mens & Tourwe, 2004) (Mens & Tourwe, 2004)
The refactoring process is still an issue (ACM October 2017) The refactoring process is still an issue (ACM October 2017)
The authors propose informal optimization models for the Refactoring The authors propose informal optimization models for the Refactoring Detection Problem ( Detection Problem (RDP RDP), making the approaches difficult to compare and ), making the approaches difficult to compare and reproduce reproduce
A A classical perspective classical perspective of the Refactoring Detection Problem
A A proposed perspective proposed perspective of the Refactoring Detection Problem
The Artificial Refactoring Generation (ARGen) is... The Artificial Refactoring Generation (ARGen) is...
Everything starts with a very precise statement of metrics Everything starts with a very precise statement of metrics
[Theoretical Refactoring] Software Formalization [Theoretical Refactoring] Software Formalization
System Under Analysis (SUA) is an information system or program System Under Analysis (SUA) is an information system or program composed of classes, methods and attributes composed of classes, methods and attributes
A single class is a Cartesian product represented by A single class is a Cartesian product represented by
α ∗ ∗
A single A single Class Class is a Cartesian product represented by is a Cartesian product represented by
α ∗ ∗
identifier fld(s) mtd(s)
∗
n
∗
n
What about What about Methods Methods and and Fields Fields?
δ
A A Refactoring Refactoring is a math function that maps from a Cartesian set to a is a math function that maps from a Cartesian set to a Code Code Modification Modification
Specific Refactoring Operation
δ
How to create the parameters? How to create the parameters?
δ
From From Source Source to to Target Target Classes Classes
δ s s s t
j α
A Quality Metric is a Function that receives a class and return a real value A Quality Metric is a Function that receives a class and return a real value
j α
We can compute any specific metric (e.g., LOC, CYCLO, LCOM2) We can compute any specific metric (e.g., LOC, CYCLO, LCOM2)
α j
α 1 α 2 α j α
[Theoretical Refactoring] Software Formalization [Theoretical Refactoring] Software Formalization
The Refactoring Impact Prediction is a technique to estimate the value of a The Refactoring Impact Prediction is a technique to estimate the value of a software metric after performing a refactoring operation (Chaparro, 2014) software metric after performing a refactoring operation (Chaparro, 2014)
δ,j α
α
Lines of Code metric impacted by Move Method Refactoring Operation Lines of Code metric impacted by Move Method Refactoring Operation
p s b s k
p t b t k
We use the concept of a typical metric, which is represent the actual value, We use the concept of a typical metric, which is represent the actual value, and the impacted metric, which is an estimation, after accounting the and the impacted metric, which is an estimation, after accounting the refactoring operation refactoring operation
j α
α
Actual Actual Metrics Metrics Forecasted Forecasted Metrics Metrics
[Theoretical Refactoring] Combinatorial Optimization [Theoretical Refactoring] Combinatorial Optimization
The search space has actionable (can perform the refactoring operation) and The search space has actionable (can perform the refactoring operation) and feasible regions (fulfill the metric's constraints) feasible regions (fulfill the metric's constraints)
e.g. a system with 10 classes, 10 attributes and 10 methods would have a size e.g. a system with 10 classes, 10 attributes and 10 methods would have a size
2 r
ARGen search space ARGen search space
The objective function is a rate that compares a Predicted from Actual System The objective function is a rate that compares a Predicted from Actual System in terms of software metrics in terms of software metrics
Set of refactoring operations Set of refactoring operations
Obj(Φ) = + ρ(Φ) w
j=1
∑
J
(
j max(Υ (η )) − min(Υ (η )) H j H j
Υ (η ) − min(Υ (η ))
H j H j
) w
j=1
∑
J
(
j max(Υ (Φ )) − min(Υ (Φ )) H ~ j H ~ j
Υ (Φ ) − min(Υ (Φ ))
H ~ j H ~ j
)
Min-max normalization Min-max normalization
Obj(Φ) = + ρ(Φ) w
j=1
∑
J
(
j max(Υ (η )) − min(Υ (η )) H j H j
Υ (η ) − min(Υ (η ))
H j H j
) w
j=1
∑
J
(
j max(Υ (Φ )) − min(Υ (Φ )) H ~ j H ~ j
Υ (Φ ) − min(Υ (Φ ))
H ~ j H ~ j
)
Developers' information Developers' information
Obj(Φ) = + ρ(Φ) w
j=1
∑
J
(
j max(Υ (η )) − min(Υ (η )) H j H j
Υ (η ) − min(Υ (η ))
H j H j
) w
j=1
∑
J
(
j max(Υ (Φ )) − min(Υ (Φ )) H ~ j H ~ j
Υ (Φ ) − min(Υ (Φ ))
H ~ j H ~ j
)
Actual Metrics Actual Metrics
Obj(Φ) = + ρ(Φ) w
j=1
∑
J
(
j max(Υ (η )) − min(Υ (η )) H j H j
Υ (η ) − min(Υ (η ))
H j H j
) w
j=1
∑
J
(
j max(Υ (Φ )) − min(Υ (Φ )) H ~ j H ~ j
Υ (Φ ) − min(Υ (Φ ))
H ~ j H ~ j
)
Estimated Metrics Estimated Metrics
Obj(Φ) = + ρ(Φ) w
j=1
∑
J
(
j max(Υ (η )) − min(Υ (η )) H j H j
Υ (η ) − min(Υ (η ))
H j H j
) w
j=1
∑
J
(
j max(Υ (Φ )) − min(Υ (Φ )) H ~ j H ~ j
Υ (Φ ) − min(Υ (Φ ))
H ~ j H ~ j
)
Penalization Penalization
Obj(Φ) = + ρ(Φ) w
j=1
∑
J
(
j max(Υ (η )) − min(Υ (η )) H j H j
Υ (η ) − min(Υ (η ))
H j H j
) w
j=1
∑
J
(
j max(Υ (Φ )) − min(Υ (Φ )) H ~ j H ~ j
Υ (Φ ) − min(Υ (Φ ))
H ~ j H ~ j
)
The refactoring repair functions account a catalog of 10 constraints based on The refactoring repair functions account a catalog of 10 constraints based on
We design 6 genetic operators for the Hybrid Optimization employed We design 6 genetic operators for the Hybrid Optimization employed
[Empirical Refactoring] Computational Technique Design [Empirical Refactoring] Computational Technique Design
Building a metaphor of the system
Compute "Actual Metrics"
Configure Individuals given Fowler's Catalog
Use unalcol
Use Estimation (RIPE) and Compute Fitness
Report in a Json File
Preliminary Experiment: Do-ability using a Shapiro-Wilk Test (non-normal Preliminary Experiment: Do-ability using a Shapiro-Wilk Test (non-normal distribution) distribution)
Algorithm CCODEC[2000] ACRA[60000] Hill Climbing 0.0049 0.0015 Simulated A. 0.0222 0.0157 HaEa 0.0144 0.0340
Large Evaluations ACRA[60000] Large Evaluations ACRA[60000]
Hill Climbing
Large Evaluations ACRA[60000] Large Evaluations ACRA[60000]
Simulated Annealing
Large Evaluations ACRA[60000] Large Evaluations ACRA[60000]
Large Evaluations ACRA[60000] for HaEa Large Evaluations ACRA[60000] for HaEa
Large Evaluations ACRA[60000] for HaEa Large Evaluations ACRA[60000] for HaEa
Hybrid Evolutionary Approach has lower values results in large iterations (0.95 +/- 0.003 [60000]) Evolutionary Algorithms seems to work better than baseline (p-value <= 0.0002) The main inconvenient during execution was latency, we had to redeploy and test with new parameters
Unified Mathematical Approximation A definition, a development, and an evaluation of ARGen Performance and complexity validation
The refactoring consistency metric is based on Archipelago (Zarras, et al. The refactoring consistency metric is based on Archipelago (Zarras, et al. 2015) 2015)
Λ λ
π
Involving self-organization and artificial minds Involving self-organization and artificial minds