IJSRSET19625 | Received : 01 March 2019 | Accepted : 10 March 2019 | March-April -2019 [ 6 (2) : 59-67]
© 2019 IJSRSET | Volume 6 | Issue 2 | Print ISSN: 2395-1990 | Online ISSN : 2394-4099
Themed Section : Engineering and Technology
DOI : https://doi.org/10.32628/IJSRSET19625
59
A Novel Based Recommended System Regularized with User Trust and Item
Rating Prediction
Anusha Viswanadapalli
1
, Praveen Kumar Nelapati
2
1
Assistant Professor, Department of Computer Science and Engineering, SRI Mittapalli Institute of Technology,
JNTU Kakinada, Andhra Pradesh, India
2
Assistant Professor, Department of Computer Science and Engineering, NRI Institute of Technology, JNTU
Kakinada, Andhra Pradesh, India
ABSTRACT
Singular Value Decomposition (SVD) is trust-based matrix factorization technique for recommendations is
proposed. Trust SVD integrates multiple information sources into the recommendation model to reduce the
data sparsity and cold start problems and their deterioration of recommendation performance. An analysis of
social trust data from four real-world data sets suggests that both the explicit and the implicit influence of both
ratings and trust should be taken into consideration in a recommendation model. Trust SVD therefore builds on
top of a state-of-the-art recommendation algorithm, SVD++ uses the explicit and implicit influence of rated
items, by further incorporating both the explicit and implicit influence of trusted and trusting users on the
guess of items for an active user. The proposed technique extends SVD++ with social trust information.
Experimental results on the four data sets demonstrate that Trust SVD achieves accuracy than other
recommendation techniques.
Keywords : Data Mining, Recommender Systems, Rating Prediction, Explicit And Implicit Influence.
I. INTRODUCTION
A Novel trust-based recommendation model, which is
regularized with user trust and item rating is Trust
SVD. Our method is novel for its consideration of
both the explicit (rating based on social circle) and
implicit influence (self-rating) of item ratings and of
the user trust. In addition, a weighted regularization
technique is used to avoid over-fitting for model
learning. This trust-based matrix factorization model
incorporates both rating and trust information for
rating prediction. Trust information is very sparse, yet
complementary to the information. Thus, focusing
too much on either one kind of information achieves
only marginal gains in predictive correctness. Also
users are strongly correlated with their trust
neighbors and have a weakly positive correlation with
their trust-alike neighbors (e.g., friends). These
observations are motivated to consider both explicit
and implicit influence of ratings and of trust in a
trust-based model. A weighted λ- regularization
technique was used to regularize the user- and item
specific latent feature vectors. This guarantees that
the user-specific vectors can be learned from their
trust information even if a few or no ratings are given.
So data sparsity and cold start issues for
recommendation can be solved. Trust SVD can
outperform both trust and ratings based methods in
the predictive accuracy. Recommender systems
employ from a specific type of information filtering
system technique that attempts to recommend
information items (movies, TV program/show/episode,
video on demand, web pages, books, news, music,
images, scientific literature etc.) or social elements