| Abstract: | The widespread of the internet caused a massive increase in its users, which helped
the online services to conquer a larger market share, and as a result, most of
our activities depend on those services. But also, it becomes easy to falsify and
fake a lot of products, which allow the creation of massive similar information
and contents of different types for the user. It results in a variety of challenges,
such as the way to choose the correct product or how to find the best possible
deal through ever-increasing data. Online services such as Netflix, Spotify, and
many other e-commerce companies who deliver on-demand services to users, have
introduced help mechanisms and created robust systems to support consumers
facing those challenges by providing suggestions based on their preferences. In
order to apply these strategies, it is essential to identify the interests and tastes of
users to recommend suitable articles. However, due to the changing user habits and
evolving trends, making relevant recommendations becomes more complicated and
a difficult task to accomplish, which requires tending towards intelligent behavior to
make recommendation engines able to be scalable to billions of users with multiple
preferences, to propose adequate elements, and to diversify recommendations.
It is thus important to implement data mining methods and machine learning
algorithms in recommender systems to identify user’s needs and extract as much
knowledge as possible from existing information to create models capable of
making suitable predictions. In this report, we will present the state of the art
of existing research in the field of the usage of data mining techniques in the
recommendation systems. Also, we will study different types of recommender
systems, data mining techniques, and machine learning algorithms which makes
it possible to make a comparison between the works analyzed then synthesis to
acquire sufficient knowledge in this field of study. |
