Journal of Computer Sciences and Applications, 2017, Vol. 5, No. 2, 76-82
Available online at http://pubs.sciepub.com/jcsa/5/2/4
©Science and Education Publishing
DOI:10.12691/jcsa-5-2-4
Efficient and Scalable Matrix Factorization Transfer
with Review Helpfulness for Massive Data Processing
Aboagye Emelia Opoku
1,2,*
, Jianbin Gao
3
, Dagadu Joshua Caleb
4
, Qi Xia
5,6
1
School of Computer Science and Engineering, University of Electronic Science and Technology of China, Chengdu, China
2
Kumasi Technical University, Kumasi, Ghana
3
School of Resource and Environment, University of Electronic, Science and Technology of China Chengdu, Chengdu, China
4
Computer Science Department, University of Electronic Science and Technology, Chengdu, China
5
School of Computer Science and Engineering, University of Electronic Science and Technology of China Chengdu, Chengdu, China
6
Center for Cyber Security, University of Electronic Science and Technology of China, Chengdu, Sichuan, China
*Corresponding author: eoaboagye@yahoo.co.uk
Abstract We explore the sparsity problem associated with recommendation system through the concept of
transfer learning (TL) which are normally caused by missing and noisy ratings and or review helpfulness. TL is a
machine learning (ML) method which aims to extract knowledge gained in a source task/domain and use it to
facilitate the learning of a target predictive function in a different domain. The creation and transfer of knowledge
are a basis for competitive advantage. One of the challenges prevailing in this era of big data is scalable algorithms
that process the massive data in reducing computational complexity. In the RS field, one of the inherent problems
researchers always try to solve is data sparsity. The data associated with rating scores and helpfulness of review
scores are always sparse presenting sparsity problems in recommendation systems (RSs). Meanwhile, review
helpfulness votes helps facilitate consumer purchase decision-making processes. We use online review helpfulness
votes as an auxiliary in formation source and design a matrix transfer framework to address the sparsity problem.
We model our Homogenous Fusion Transfer Learning approach based on Matrix Factorization HMT with review
helpfulness to solve sparsity problem of recommender systems and to enhance predictive performance within the
same domain. Our experiments show that, our framework Efficient Matrix Transfer Learning (HMT) is scalable,
computationally less expensive and solves the sparsity problem of recommendations in the e-commerce industry.
Keywords: fusion, transfer learning, sparsity, helpfulness
Cite This Article: Aboagye Emelia Opoku, Jianbin Gao, Dagadu Joshua Caleb, and Qi Xia, “Efficient and
Scalable Matrix Factorization Transfer with Review Helpfulness for Massive Data Processing.” Journal of
Computer Sciences and Applications, vol. 5, no. 2 (2017): 76-82. doi: 10.12691/jcsa-5-2-4.
1. Introduction
Massive data analytics is aimed at making sense of data
by applying efficient and scalable algorithms on big data
for its analysis, learning, modelling, visualization and
understanding. this requires the design of efficient and
effective algorithms and systems to integrate the data
and uncover the hidden values from it. It also requires
methodologies and algorithms for automatic or
mixed-initiative knowledge discovery and learning, data
transformation and modelling, predictions and
explanations of the data [1,2]. Online purchasing at
various e-commerce sites like amazon, Netflix are
crowded with chunks of data. Consumer purchasing
decision making of such online products are immensely
affected by their reviews and its associated information in
such massive data environments [3,4], in such a way that,
information overload has now become a constraint to
consumers daily [5]. In this paper we focus on the sparsity
problem that arises with review helpfulness giving rise to
sparse matrix. the sparsity problem originates when
available data are insufficient for identifying similar users
or we can say neighbours and it is a major issue since it
limits the quality of recommendations and the
applicability of collaborative filtering in general. The main
objective of our work is to develop an effective approach
that provides high quality recommendations even when
sufficient data are unavailable [6]. We use matrix
factorization integration to reduce the sparsity using
homogenous transfer learning paradigm. This on one hand
must provide valuable product related information for
consumers which in the long run affect the quality of
recommendation system algorithms [7]. Consequently,
researchers from diverse fields have explored the most
efficient and effective ways of predicting quality
recommendations. One of such paradigms in recent times
has been transfer learning [8,9,10] from a psychological
point of view, transfer learning is the study of the
dependency of human conduct, learning, or performance
on prior experience [11]. For instance, if one is good at
coding in object oriented programming coding, may learn
python programming language principally. Our main
motivation for application of transfer learning in the
recommender system field is the need for machine