Thesis (MSc) - Cyprus International University. Institute of Graduate Studies and Research Information Systems Engineering Department

Includes bibliography (sheets 51-54)

ABSTRACT
Accessibility to the web is a per-requisite for every e-commerce website to maintain
and retain its business in today’s business world. Web users devote interests in surfing
the net which has made abundant information available for data mining. This data can
be examined to reveal a lot of patterns that would help users to find their choice of
pages or products as the case may be. For this to be successful, enhanced mining
technique background are required to fully extract, comprehend and interpret these
data. To assist users of e-commerce websites to finding products and resources of their
choice, an automated approach is required to recommend products or resources to
users to lessen the time spent on shopping. Well-known companies have employed
recommendation system to their websites to easy the stress and time spent by users
which as well have added more benefits to the enterprise. In this research, we have
discussed collaborative, content-based and hybrid methods of recommendation
system. Challenges of the existing systems and their various advantages were also
elaborated. The objective of the research is to recommend relevant products to
customers as irrelevant products serves as disruption during shopping. In our hybrid
method of recommendation, we have recommended five top products to a user using
dataset from Amazon electronic product review and the performance of these models
were also evaluated.
Keywords: Collaborative filtering, Hybrid filtering, Product-based recommendation,
Singular Value Decomposition.