Advances in Science and Technology – Research Journal,
Год журнала:
2023,
Номер
17(4), С. 162 - 167
Опубликована: Авг. 7, 2023
The
practice
of
code
review
is
crucial
in
software
development
to
improve
quality
and
promote
knowledge
exchange
among
team
members.It
requires
identifying
qualified
reviewers
with
the
necessary
expertise
experience
thoroughly
examine
modifications
suggested
a
pull
request
efficiency
process.However,
it
can
be
costly
time-consuming
for
maintainers
manually
assign
suitable
each
large-scale
projects.To
address
this
challenge,
various
techniques,
including
machine
learning,
heuristic-based
algorithms,
social
network
analysis,
have
been
employed
suggest
requests
automatically.The
primary
challenge
recommending
verifying
whether
accurate.While
there
attempts
replicate
previous
recommendation
processes
or
propose
new
methods,
evaluating
correctness
reviews
remains
area
research.New
approaches
are
emerging
assess
reviews,
but
further
research
needed
develop
more
reliable
methods
that
applied
contexts.This
study
investigates
an
automated
evaluation
accuracy
its
impact
on
possible.One
possible
approach
use
pre-trained
language
model
like
ChatGPT3
extract
key
information
from
text.Another
method
NLP
techniques
automatically
generate
annotations
text,
which
could
then
used
train
learning
predict
accurately.Automated
mechanisms
potential
positively
both
open
source
industry
projects
by
increasing
transparency
accountability
process
improving
overall
project
outcomes.Therefore,
developing
implementing
effective
systems
areas
significant
benefits.
Business Horizons,
Год журнала:
2024,
Номер
unknown
Опубликована: Май 1, 2024
Generative
AI
(artificial
intelligence)
technologies
using
LLMs
(large
language
models),
such
as
ChatGPT
and
GitHub
Copilot,
with
the
ability
to
create
code,
have
potential
change
software
development
landscape.
Will
this
process
be
incremental,
developers
learning
generative
skills
supplement
their
existing
skills,
or
will
more
destructive,
loss
of
large
numbers
jobs
a
radical
in
responsibilities
remaining
developers?
Given
rapid
growth
capabilities,
it
is
impossible
provide
crystal
ball,
but
article
aims
give
insight
into
adoption
development.
The
gives
an
overview
industry
job
functions
developers.
A
literature
review
combined
content
analysis
online
comments
from
how
implemented
changing
are
responding
these
changes.
ties
academic
developer
insights
together
recommendations
for
developers,
describes
CMM
(capability
maturity
model)
framework
assessing
improving
LLM
usage.
Electronics,
Год журнала:
2025,
Номер
14(2), С. 382 - 382
Опубликована: Янв. 19, 2025
Human–machine
pair
inspection
refers
to
a
technique
that
supports
programmers
and
machines
working
together
as
“pair”
in
source
code
tasks.
The
machine
provides
guidance,
while
the
programmer
performs
based
on
this
guidance.
Although
are
often
best
suited
inspect
their
own
due
familiarity,
overconfidence
may
lead
them
overlook
important
details.
This
study
introduces
novel
mutation-based
human–machine
method,
which
is
designed
direct
programmer’s
attention
specific
components
by
applying
targeted
mutations.
We
assess
effectiveness
of
inspections
analyzing
corrections
these
Our
approach
involves
defining
mutation
operators
for
each
keyword
program
historical
defects,
developing
rules
keywords
strategy
automatically
generating
mutants,
designing
comparison
quantitatively
evaluate
quality.
Through
controlled
experiment,
we
demonstrate
aiding
during
process.
ACM Transactions on Software Engineering and Methodology,
Год журнала:
2025,
Номер
unknown
Опубликована: Фев. 28, 2025
Accountability
is
an
innate
part
of
social
systems.
It
maintains
stability
and
ensures
positive
pressure
on
individuals’
decision-making.
As
actors
in
a
system,
software
developers
are
accountable
to
their
team
organization
for
decisions.
However,
the
drivers
accountability
how
it
changes
behavior
development
less
understood.
In
this
study,
we
look
at
aspects
code
review
affect
engineers’
sense
quality.
Since
engineering
(SE)
increasingly
involving
Large
Language
Models
(LLM)
assistance,
also
evaluate
impact
when
introducing
LLM-assisted
reviews.
We
carried
out
two-phased
sequential
qualitative
study
(
\(\textbf{interviews}\rightarrow\textbf{focus
groups}\)
).
Phase
I
(16
interviews),
sought
investigate
intrinsic
engineers
influencing
quality,
relying
self-reported
claims.
II,
tested
these
traits
more
natural
setting
by
simulating
traditional
peer-led
reviews
with
focus
groups
then
sessions.
found
that
there
four
key
quality:
personal
standards
,
professional
integrity
pride
quality
maintaining
one’s
reputation
.
review,
observed
transition
from
individual
collective
initiated.
introduction
disrupts
process,
challenging
reciprocity
taking
place
evaluations,
i.e.,
one
cannot
be
LLM.
Our
findings
imply
AI
into
SE
must
preserve
mechanisms.
IEEE Software,
Год журнала:
2024,
Номер
41(6), С. 38 - 45
Опубликована: Июль 18, 2024
In
this
study,
we
investigate
what
has
been
discussed
about
generative
AI
in
the
code
review
context
by
performing
a
gray
literature
review.
We
analyzed
42
documents
and
found
insights
from
practice
proposals
of
solutions
using
models.
ACM Transactions on Software Engineering and Methodology,
Год журнала:
2024,
Номер
unknown
Опубликована: Дек. 19, 2024
As
our
lives,
businesses,
and
indeed
world
economy
become
increasingly
reliant
on
the
secure
operation
of
many
interconnected
software
systems,
engineering
research
community
is
faced
with
unprecedented
challenges,
but
also
exciting
new
opportunities.
In
this
roadmap
paper,
we
outline
vision
Software
Security
Analysis
for
systems
future.
Given
recent
advances
in
generative
AI,
need
methods
to
assess
maximize
security
code
co-written
by
machines.
heterogeneous,
practical
approaches
that
work
even
if
some
functions
are
automatically
generated,
e.g.,
deep
neural
networks.
depend
evermore
supply
chain,
tools
scale
an
entire
ecosystem.
What
kind
vulnerabilities
exist
future
how
do
detect
them?
When
all
shallow
bugs
found,
discover
hidden
deeply
system?
Assuming
cannot
find
flaws,
can
nevertheless
protect
To
answer
these
questions,
start
a
survey
security,
then
discuss
open
challenges
opportunities,
conclude
long-term
perspective
field.
