Concurrency and Computation Practice and Experience,
Год журнала:
2024,
Номер
unknown
Опубликована: Дек. 10, 2024
ABSTRACT
The
algorithm
for
multimodal
image‐text
retrieval
aims
to
overcome
the
differences
between
visual
and
textual
data,
enabling
efficient
accurate
recognition
images
text.
Since
manually
labeled
data
are
usually
expensive,
many
researchers
attempted
use
low‐quality
obtained
through
network
batch
operations.
This
presents
a
challenge
model's
generalization
performance
prediction
accuracy.
To
address
this
issue,
we
construct
system
of
based
on
fusion
pre‐trained
models.
Firstly,
enhance
diversity
original
using
MixGen
improve
performance.
Next,
employ
Chinese‐CLIP
as
most
suitable
foundational
model
comparative
experiments
among
three
different
Finally,
comprehensive
ensemble
with
base
models
specific
characteristics
tasks,
which
includes
prediction‐based
text‐to‐image
task
feature‐based
image‐to‐text
task.
Extensive
show
that
our
outperforms
state‐of‐the‐art
single
foundation
in
generalization,
especially
pairs
small
datasets
Chinese
context.
Concurrency and Computation Practice and Experience,
Год журнала:
2025,
Номер
37(4-5)
Опубликована: Фев. 10, 2025
ABSTRACT
In
digital‐intelligent
warehouses,
the
heavy
handling
tasks,
complex
algorithms
with
high
computational
demands,
and
vast
solution
spaces
pose
significant
challenges
to
achieving
stable,
efficient,
balanced
operation
of
multiple
Autonomous
Mobile
Robots
(AMRs)
for
automated
cargo
handling.
This
paper
focuses
on
a
virtual
smart
warehouse
environment
employs
Python
software
conduct
simulation
experiments
multi‐AMR
task
allocation.
The
simulated
comprises
three
idle
AMRs
16
points
that
require
transportation.
experimental
simulations
demonstrate
improved
genetic
algorithm
can
find
global
optimal
relatively
low
cost,
meeting
fast
response
requirements
in
real‐world
operations.
It
enables
stable
operation,
efficiency,
allocation
AMRs.
results
validate
reliability
proposed
method,
effectively
addressing
issues
path
planning
warehouses.
Concurrency and Computation Practice and Experience,
Год журнала:
2025,
Номер
37(4-5)
Опубликована: Фев. 19, 2025
ABSTRACT
The
distributed
electric
drive
vehicle
is
a
highly
nonlinear
and
time‐varying
system.
To
address
the
issue
of
slip
control
under
varying
driving
forces
road
surface
coefficients,
novel
strategy
proposed,
which
considers
axle
load
transfer
during
acceleration.
employs
an
improved
PSO
algorithm
to
obtain
optimal
parameters
for
BP
neural
network,
uses
network
forward
propagation
calculate
PID
in
real‐time,
adjusts
weight
matrix
through
backward
achieve
real‐time
adaptive
slip.
Experimental
results
indicate
that
this
improves
ITAE
index
by
13.6%
response
time
74.8%
compared
anti‐saturation
PID.
Sustainability,
Год журнала:
2025,
Номер
17(5), С. 1964 - 1964
Опубликована: Фев. 25, 2025
The
development
of
the
shipping
and
port
industry
has
significantly
driven
both
Chinese
global
economies.
However,
environmental
impact
their
carbon
footprint
cannot
be
overlooked.
Recently,
made
strides
in
implementing
various
protection
measures,
with
promotion
adoption
shore
power
technology
inland
areas
emerging
as
promising
avenues
for
sustainable
development.
Despite
widespread
establishment
facilities
river
areas,
utilization
rate
remains
low.
This
is
primarily
due
to
minimal
benefits
that
ships
gain
from
using
often
fail
meet
expectations.
traditional
income
function
based
on
rational
agent
theory
inadequate,
leading
existing
incentive
measures
do
not
effectively
encourage
usage
by
ships.
Therefore,
it
essential
analyze
this
issue
perspective
behavioral
economics.
paper
introduces
reference
point
effect
prospect
designs
an
mechanism
game
principles.
goal
explore
how
government
subsidies
penalties
can
adjusted
create
effective
incentives
usage,
ultimately
seeking
improve
among
enhance
overall
quality
ports.
Concurrency and Computation Practice and Experience,
Год журнала:
2025,
Номер
37(12-14)
Опубликована: Май 16, 2025
ABSTRACT
In
recent
years,
there
has
been
a
notable
increase
in
food
safety
incidents,
which
raised
considerable
public
concern.
Optimizing
supervision
and
enhancing
trust
have
become
urgent
issues
to
be
addressed.
This
study
specifically
examines
the
“tanker
mixed
with
edible
oil”
incident
employs
variety
of
methodologies,
including
text
analysis
time
series
modeling,
conduct
comprehensive
sentiment,
The
findings
provide
scientific
foundation
for
regulatory
oversight.
Relevant
data
were
gathered
via
Python,
opinion
trends
forecast
ARIMA
model,
an
in‐depth
thematic
characteristics
associated
each
phase
development
was
conducted
by
integrating
LDA
topic
modeling
techniques.
Meanwhile,
this
social
network
construct
interactive
among
users
identify
key
nodes
pathways
involved
dissemination
opinion.
Through
simulation
analysis,
following
conclusions
are
drawn:
(1)
cooking
exhibited
pronounced
trend
negative
sentiment
that
intensified
over
time.
(2)
reveals
concern
regarding
disarray
transportation
insufficient
oversight,
highlighting
shift
public's
focus.
(3)
Social
emphasizes
crucial
roles
played
official
media
individual
leaders
(KOLs)
shaping
opinion,
illustrating
how
these
entities
influence
direction
through
their
relationships.
empirical
incident,
paper
verifies
effectiveness
adopted
method,
providing
important
reference
risk
prevention
control
policy‐making.
Internet Research,
Год журнала:
2025,
Номер
unknown
Опубликована: Май 19, 2025
Purpose
This
study
investigates
the
impact
of
social
media
posts
on
panic
buying
behavior
while
also
examining
moderating
role
posts’
attention
level
and
exploring
mediating
effects
perceived
credibility,
risk
scarcity.
Design/methodology/approach
Building
upon
elaboration
likelihood
model
scarcity
theory,
this
examines
impacts
numerical
presentation
posting
account
type
explores
their
underlying
mechanisms.
To
test
our
hypotheses,
we
conducted
two
online
randomized
experiments
employed
t
-tests
regression
analysis
as
main
analytical
methods.
Findings
Our
findings
highlighted
significance
incorporating
precise
information
utilizing
institutional
accounts
to
enhance
users’
perceptions
thereby
inducing
behavior.
Moreover,
from
were
found
amplify
risks.
Additionally,
discovered
that
both
negatively
moderated
by
level.
Originality/value
While
previous
research
has
acknowledged
behavior,
it
overlooked
identification
specific
attributes
within
these
may
exert
influence
well
neglected
consider
user
attitudes.
contributes
existing
literature
elucidating
mechanisms
through
which
three
distinct
characteristics
providing
practical
insights
for
effectively
managing
content
mitigating
buying.
Concurrency and Computation Practice and Experience,
Год журнала:
2025,
Номер
37(12-14)
Опубликована: Май 7, 2025
ABSTRACT
Carbon
emissions
are
a
significant
contributor
to
global
warming.
As
one
of
the
largest
carbon
emitters
in
world,
China
is
committed
establishing
emission
trading
market
address
challenges
posed
by
climate
change.
The
price
fundamental
component
financial
market.
Accurately
predicting
it
can
improve
environmental
quality,
reduce
energy
demand,
and
promote
economic
growth.
This
study
uses
data
from
Guangdong
as
case
employs
hybrid
model
that
integrates
Convolutional
Neural
Networks
(CNN)
Long
Short‐Term
Memory
(LSTM)
networks
for
forecasting.
findings
indicate
that:
(1)
CNN–LSTM
exhibits
optimal
predictive
performance
when
sliding
window
set
size
5
on
basis
previous
data.
(2)
By
incorporating
indicator
features
pilot
dataset
while
maintaining
5,
achieves
superior
accuracy,
evidenced
Goodness
Fit
(
R
2
)
0.8622
mean
absolute
error
(MAE)
0.0228,
resulting
most
favorable
comprehensive
evaluation
index.
(3)
integration
one‐dimensional
convolutional
layers
with
LSTM
effectively
leverages
strengths
CNNs
local
feature
extraction
capabilities
LSTMs
modeling
time
series
approach
leads
substantial
improvement
compared
alternative
models
such
Support
Vector
Machine
(SVM),
Recurrent
Network
(RNN),
LSTM.