Search-Based Software Engineering: Challenges, Opportunities and Applications - Tutorial @GECOO 2020

hello I'm glad to present this tutorial

today on search method software

engineering challenges opportunities and

recent applications my name is Adina I'm

from ETS Montreal University of Quebec

in Canada so first of all I would like

to thank professors Mark Harmon and

Marvin Santini for the inspiration to

prepare part of this tutorial and I'm

also grateful for all my students

co-authors and collaborators this work

have never been possible without their

support and collaboration so this

tutorial today will cover different

aspects I will start by presenting the

biological basis behind science and

engineering and then I will describe

what is SBC and how we apply it in

practice then to better understand this

message I will present three

applications of SPS see what under

commendation of refactoring changes the

second application is about improving

the web services design and the third

application is about reviewing code

changes in modern code review the

wraptor I will present the hands-on

activity using an existing open source

framework called

my framework and simplified case study

on software migration and finally I'll

be presenting and discussing the

challenges and future work with search

based software engineering so let's

start with the philosophical basis

behind science and engineering so

science is more about using

experimentation and analysis to explore

the world around us so we can take

problem we characterize it quantify and

model the problem this is mainly the

role of scientist on the other side in

engineering we take this information

provided by scientists and use it to

design and develop products that are

gonna allow us to quality human life to

continue

Engineering offers humanity the hope

because science gives us a state of

information but no concrete solution so

the scientist mainly is looking at how

things are theoretically done while the

engineer is looking at how things are

practically done

the scientist Murli explores what exists

but the engineer is more creating what

has never been existed before

so as analogy a computer scientist we

look at the fundamental principles

behind computation and how it works well

the idea for an engineer is to complete

specifications or meet designs but

within a set of constraints and

limitations so how would you consider

yourself yourself

are you more engineer or scientist so

the the synergy or the innate

relationship between both science and

engineering is on the basis of search

based software engineering so search

based software engineering inert from

both fields

computer science and software

engineering so let's move to the first

part 2 of this tutorial about what is as

BSC so before delving into the details

of especially let us see what are the

common work that engineers talk about

all the time so we check it from

different engineering fields they mostly

talk about optimized analyzed solution

estimate tolerance constraint model we

found that the dimensional optimized

wise so one in breach and one in

American so good so as busy is mainly

about optimization is in the core of

SBAC the term SVC was coined first by

Mark Harmon in 2001 and SPC is about the

use of intelligent search techniques to

explore large search spaces guided by

what we call Fitness function that

captures the properties of a desirable

solutions that we need we seek to have

and there exist several ready-made

algorithms such as the Genetic

Programming harmony search particle

swarm optimization and simulated

annealing all these algorithms will be

used to solve any optimization

problem including in software

engineering and other fields so these

algorithms fall under the evolutionary

search algorithms as part of machine

learning which are all artificial

intelligence techniques to be used for

robotics originally and other fields so

SBC is mainly an intersection between

two fields the optimization and software

engineering in software engineering we

have multiple problems that need to be

solved like automated test generation

code refactoring software project

management requirements engineering

fault

localization and program repair the

field of search based optimization

provides us with a set of ready-made

algorithms that are ready to use from

single objective to monthly or many

objective I graded such as the genetic

algorithm the non dominated sort and

genetic algorithm etc and the

application of serrated optimization to

software engineering would create the

field of SBAC we found search basic

testing which is an active field in the

software engineering community we have

also search based refactoring search

based software management search basic

requirements engineering etc so let's

see how or what is SBC in in a natural

so we mainly start from software

engineering problem then we try to use a

solution representation or a specific

encoding for the solution we can

represent the solution as for example a

vector or a 3-2 to fix a given problem

then we need to define what we call

Fitness function that evaluates how good

is our represented solution then we need

to define change operators that can

refine this the candidate solution to

make it better and that can involve over

an evolutionary process to make the

solution better and fitter

so the field of search basis of fringing

is growing

very fast overtime so as we can see the

trend of the use of SDSC until mid-2012

we have over 1,600 authors from nearly

300 institutions spread over more than

40 countries so this is realistic

scenario and the philosophy behind this

busy pushed a lot of researchers and

practitioners to use SBAC to solve their

problems and today we are dedicated

tracks in the top conferences like in XE

or in FSC or also in top conferences in

evolutionary computation like geek oh

there is a dedicated track for SBC

there are also some dedicated

international conferences back or

symposiums like SSB se or sbst we found

also several publications in the top

journals fat asian Tosa in the recent

years so maybe the question is why this

is growing that fast the answer is that

in like other fields we have a valuable

opportunity to check many possible

solutions for a given a problem and then

select the best one in physical

engineering this could be limited and

costly as we need to try different

solutions and experiments and the cost

can limit the application of these

simulations so however in software

engineering for example to fix a bug we

can change the structure of a program

and try different patches to find the

Bell and then find the best one so there

is almost no cost for such experiment we

can modify the code as much as we can

this makes software engineering more

attractive and more applicable in

practice so for example let us see a

real world example of space rocket so

the goal is to find the optimal

structure that maximizes the compression

and minimize the forth conception

consumption this is an extremely costly

experiment we can try cannot write

physically on the space so people

usually

use simulations but there is a risk that

simulation is not as similar to the

reality software engineering is more

attractive in this case for example to

find an optimal design that maximize

cohesion and minimize coupling this is

possible so we can modify the code see