Objective
To
address
the
complexities
of
distinguishing
truth
from
falsehood
in
context
COVID-19
infodemic,
this
paper
focuses
on
utilizing
deep
learning
models
for
infodemic
ternary
classification
detection.
Methods
Eight
commonly
used
are
employed
to
categorize
collected
records
as
true,
false,
or
uncertain.
These
include
fastText,
three
based
recurrent
neural
networks,
two
convolutional
and
transformer-based
models.
Results
Precision,
recall,
F1-score
metrics
each
category,
along
with
overall
accuracy,
presented
establish
benchmark
results.
Additionally,
a
comprehensive
analysis
confusion
matrix
is
conducted
provide
insights
into
models’
performance.
Conclusion
Given
limited
availability
relatively
modest
size
tested
data
sets,
pretrained
embeddings
simpler
architectures
tend
outperform
their
more
complex
counterparts.
This
highlights
potential
efficiency
detection
underscores
need
further
research
area.
ACM Transactions on Multimedia Computing Communications and Applications,
Год журнала:
2025,
Номер
unknown
Опубликована: Янв. 10, 2025
With
the
popularization
of
social
networks,
fake
news
is
also
widely
and
rapidly
spreading,
which
poses
a
great
threat
to
Internet.
Therefore,
how
detect
automatically
efficiently
has
become
an
urgent
problem
be
solved.
However,
existing
approaches
mostly
focus
on
explicit
features
(images
text)
deep
fusions,
without
considering
potential
such
as
text
emotion
image
category.
To
find
solution
this
issue,
we
propose
Potential
Features
Fusion
Network
(PFFN),
models
at
same
time.
exploit
features,
introduce
mixture
experts
structure
process
separately,
can
best
use
relationships
between
category
detection.
Besides,
extract
fuse
them
with
features.
Finally,
establish
attention-based
feature
fusion
network
obtain
multi-modal
piece
thus
further
improve
performance.
We
make
experiments
four
public
datasets
(Weibo16,
Weibo19,
Twitter,
PolitiFact),
results
compared
baseline
demonstrate
that
our
PFFN
better
Our
code
available
https://github.com/Wang-bupt/PFFN
Engineering Technology & Applied Science Research,
Год журнала:
2024,
Номер
14(4), С. 15665 - 15675
Опубликована: Авг. 2, 2024
Today,
detecting
fake
news
has
become
challenging
as
anyone
can
interact
by
freely
sending
or
receiving
electronic
information.
Deep
learning
processes
to
detect
multimodal
have
achieved
great
success.
However,
these
methods
easily
fuse
information
from
different
modality
sources,
such
concatenation
and
element-wise
product,
without
considering
how
each
affects
the
other,
resulting
in
low
accuracy.
This
study
presents
a
focused
survey
on
use
of
deep
approaches
visual
textual
various
social
networks
2019
2024.
Several
relevant
factors
are
discussed,
including
a)
detection
stage,
which
involves
algorithms,
b)
for
analyzing
data
types,
c)
choosing
best
fusion
mechanism
combine
multiple
sources.
delves
into
existing
constraints
previous
studies
provide
future
tips
addressing
open
challenges
problems.
