| dc.description.abstract | The past few years have seen the growth of the data industry in every aspect of life. Data
Science has become so powerful and popular that it triggers most of the life motion even without
the acknowledgement of the users. Business is one of the fields that make use of data flow most
efficiently, especially in e-commerce. Among the diversity of techniques involving data learning,
a recommendation system appears to be quite well-known as it improves customer experience,
retention and sales. It is an application of Data mining in capturing user’s behavior and pattern
through mining user’s previous purchasing information. Above that, a Self-Attentive Sequential
Recommender focuses on finding the pattern from sequential dynamics and investigating the
dependencies of each item in a sequence with a mechanism called Self-Attention [1]. While
recommendation systems learn the customers’ behavior and make predictions to enhance sales,
topic modeling, which is the most common approach of Natural Language Processing, identifies
the main topics or themes from text-based content such as product description or customer review
to come up with products that are the most likely suitable for a specific user. Topic modeling,
including many approaches like Latent Dirichlet Allocation (LDA) and Latent Semantic Analysis
(LSA) is normally used along with Recommender System to improve the accuracy and relevance
of the product recommendations. In the scope of this thesis proposal, LSA can be used to boost the
performance of SASRec by combining those two techniques in one framework. | en_US |
