GenSim : GAN based Recommendation systems for personalized matrix factorization
Abstract
Accurately
modeling
user
preferences
is
crucial
for
the
success
of
modern
recommendation
systems
(RS).
Despite
recent
advances
in
generative
models
RS,
challenges
such
as
limited
data
and
complexity
human
behavior
still
persist.
These
issues
make
it
difficult
to
generate
accurate
authentic
profiles,
which
are
essential
providing
meaningful
personalized
recommendations.
In
this
paper,
we
introduce
GenSim,
a
novel
approach
that
combines
Generative
Adversarial
Networks
(GANs),
Genetic
Algorithms
(GAs),
similarity
techniques
overcome
critical
collaborative
filtering
(CF),
sparsity
intricate
user-item
interactions.
By
integrating
these
methods,
GenSim
offers
robust
scalable
framework
enhancing
RS
performance.
A
key
feature
its
focus
on
matrices,
selectively
consider
only
similar
users
or
items,
rather
than
entire
matrix.
This
targeted
refines
input
during
generation
phase,
resulting
recommendations
not
more
leading
accurate,
personalized,
efficient
Our
integrates
GAN
with
an
autoencoder-based
discriminator
optimized
generator
matrix
factorization,
incorporating
Pearson
enrich
process.
GAs
employed
two
phases:
preprocessing
refine
measures
fine-tuning
generator’s
hyperparameters
optimal
factorization.
Extensive
experiments
benchmark
datasets—MovieLens
1M,
HetRec,
LastFM—demonstrate
GenSim's
superior
performance
across
Precision,
MAP,
NDCG,
compared
state-of-the-art
methods.
improves
precision
by
40%
at
cutoff
50
MovieLens
1M
dataset,
MAP
38%
5
LastFM
previous
works
using
GANs
factorization
systems.
Research Square (Research Square), Journal Year: 2025, Volume and Issue: unknown
Published: April 25, 2025
Language: Английский