PeerJ Computer Science,
Год журнала:
2024,
Номер
10, С. e2213 - e2213
Опубликована: Авг. 1, 2024
Traditional
methods
may
be
inefficient
when
processing
large-scale
data
in
the
field
of
text
mining,
often
struggling
to
identify
and
cluster
relevant
information
accurately
efficiently.
Additionally,
capturing
nuanced
sentiment
emotional
context
within
news
is
challenging
with
conventional
techniques.
To
address
these
issues,
this
article
introduces
an
improved
bidirectional-Kmeans-long
short-term
memory
network-convolutional
neural
network
(BiK-LSTM-CNN)
model
that
incorporates
semantic
analysis
for
high-dimensional
visual
extraction
media
hotspot
mining.
The
BiK-LSTM-CNN
comprises
four
modules:
preprocessing,
clustering,
analysis,
itself.
By
combining
components,
effectively
identifies
common
features
input
data,
clusters
similar
articles,
analyzes
semantics
text.
This
comprehensive
approach
enhances
both
accuracy
efficiency
Experimental
results
demonstrate
compared
models
such
as
Transformer,
AdvLSTM,
NewRNN,
achieves
improvements
macro
by
0.50%,
0.91%,
1.34%,
respectively.
Similarly,
recall
rates
increase
0.51%,
1.24%,
1.26%,
while
F1
scores
improve
0.52%,
1.23%,
1.92%.
shows
significant
time
efficiency,
further
establishing
its
potential
a
more
effective
analyzing
International Journal of Computational and Experimental Science and Engineering,
Год журнала:
2025,
Номер
11(1)
Опубликована: Фев. 28, 2025
The
e-commerce
receives
extreme
competition
in
recent
years,
significantly
with
the
requirement
of
facing
demands
consumers
speed,
effective
and
accessibility.
distribution
systems
composes
crucial
role
assurance
faster
exact
delivery
products
from
warehouses
to
consumers.
Due
growth
globalized
e-commerce,
there
is
an
increasing
demand
for
classic
manageable
distributor
systems.
conventional
includes
stocking
shipping
directly
fails
deliveries
tracking
orders.
Hence,
distributors
requires
integrate
parameters
such
as
maintenance
records,
orders
logistics
on
time
without
extra
costs.
above
manages
issues
weather
modifications
disturbance
supply
chains
multi-channel
issues.
ML
DL
algorithms
allows
business
transferring
traditional
potential
data
driven
techniques.
examines
earlier
real
forecasting
whereas
assess
formless
feedbacks
fashions
social
media
additional
innovations.
utilization
those
enhances
ability
operations,
reduction
cost
increased
fulfilment
resulting
enlarged
sector.
Moreover,
are
fine-tuning
future
enhancement
generating
capability
modifying
iterative
market
transitions
needs
Journal of Advanced Computational Intelligence and Intelligent Informatics,
Год журнала:
2025,
Номер
29(1), С. 215 - 223
Опубликована: Янв. 19, 2025
Enhancing
the
precision
of
supply
chain
management
and
reducing
operational
costs
are
crucial
for
development
cross-border
e-commerce
market.
However,
existing
research
often
overlooks
demand
uncertainty
caused
by
seasonal
variations
challenges
handling
returns
in
logistics.
Therefore,
this
paper
proposes
a
SARIMA-CNN-BiLSTM
prediction
model
that
effectively
captures
both
nonlinear
characteristics
chains.
Additionally,
incorporating
process,
distribution
optimization
is
developed
with
objective
minimizing
total
costs.
The
solved
using
an
improved
whale
algorithm.
In
validation
real-world
data,
achieved
mean
absolute
percentage
error
reduction
6.479
7.703
compared
to
convolutional
neural
network
(CNN)
BiLSTM
models,
respectively.
Moreover,
chosen
algorithm
reduced
cost
231,310
CNY,
62,564
131,632
CNY
algorithm,
genetic
particle
swarm
optimization,
proposed
approach
provides
robust
support
enterprises
enhancing
efficiency
their
operations.
Journal of Computational Methods in Sciences and Engineering,
Год журнала:
2025,
Номер
unknown
Опубликована: Фев. 27, 2025
This
paper
introduces
an
innovative
online
resource
recommendation
system
tailored
for
English
text
and
video
content,
leveraging
the
power
of
attention
mechanisms
graph
neural
networks.
Given
exponential
growth
learning
resources,
a
crucial
challenge
lies
in
delivering
personalized
efficient
recommendations
to
users.
Our
study
strives
optimize
both
accuracy
efficiency
these
by
harnessing
synergistic
effects
GNNs.
By
collecting
analyzing
large
amount
user
behavior
data,
we
build
user-resource
interaction
graph.
not
only
contains
information
between
users
but
also
incorporates
association
providing
rich
context
subsequent
recommendations.
We
introduce
mechanism
handle
node
edge
graphs.
assessing
significance
various
nodes
edges
process,
are
able
capture
users’
interests
preferences
with
greater
precision.
According
experimental
integration
has
led
notable
improvement
system’s
accuracy,
achieving
increase
approximately
15%.
significant
enhancement
underscores
effectiveness
effectively
capturing
interests.
Additionally,
leverage
networks
model
intricate
structural
within
With
convolution
operations,
potential
relationships
resources
use
process.
Experimental
results
show
that
combined
GNN,
coverage
increased
about
20%,
more
diverse
results.
The
proposed
based
on
GNN
achieved
improvements
diversity
In
future,
will
further
explore
optimization
methods
provide
services.
