UNIVERSITI TEKNOLOGI MARA
MOVIE RECOMMENDATION SYSTEM
NAJWA SYAMIMIE BINTI HASNU
BACHELOR OF COMPUTER SCIENCE (Hons.)
COMPUTER SCIENCE
APRIL 2020
TABLE OF CONTENTS
CONTENT PAGE
LIST OF FIGURES 4
LIST OF TABLES 4
ABSTRACT 5
CHAPTER ONE: INTRODUCTION
1.1 Background of Study 6
1.2 Problem Statement 9
1.3 Objective 10
1.4 Project Scope 10
1.5 Project Significance 11
1.6 Overview of Research Framework 12
1.7 Summary 13
CHAPTER TWO: LITERATURE REVIEW
2.1 Introduction 14
2.2 Recommendation Systems 14
2.2.1 Phases of Recommendation Systems 15
2.2.2 Recommendation System Techniques 17
2.2.2.1 Content-Based Filtering Technique 18
2.2.2.2 Collaborative Filtering Technique 19
2.2.2.3 Hybrid Filtering 20
2.3 Collaborative Filtering 22
2.4 Implementation of Collaborative Filtering in various problems 24
2.5 Similar Works 27
2.6 The Implication of Literature Review 33
2.7 Conclusion 34
CHAPTER THREE: METHODOLOGY
3.1 Overview of Research Methodology Framework 35
3.1.1 Detailed Content of Research Methodology Framework 35
3.2 Preliminary Phase 38
3.2.1 Literature Study 38
3.3 Data Analysis 38
3.3.1 Data Collection 39
3.3.2 Data Pre-processing 40
3.3.2.1 Data Cleaning 41
2
3.3.2.2 Data Integration 42
3.4 Design and Implementation Phase 43
3.4.1 Design Interface 43
3.4.2 Process in Constructing Movie Recommendation System
Using Collaborative Filtering 47
3.4.3 Prototype Implementation 48
3.5 Performance Evaluation 49
3.6 Conclusion 50
CHAPTER FOUR: RESULT AND FINDINGS
4.1 Conceptual Framework 51
4.2 Program Codes for Algorithm 52
4.3 Prototype Interfaces 55
4.4 Evaluation Results 57
4.5 Discussion 58
4.6 Conclusion 58
CHAPTER FIVE: CONCLUSION AND RECOMMENDATION
5.1 Summary of Project 59
5.2 Project Contribution 59
5.3 Project Limitation 59
5.4 Project Recommendation 60
5.5 Conclusion 60
REFERENCES 61
APPENDIX 67
3
ABSTRACT
Movie Recommendation System is essentially developed for the purpose of producing
a movie recommendation list therefore, it is paired with algorithm to provide the
desired functionalities and features in the system. As known, users prefer to use system
that recommend suitable items according to their preferences so that they will not have
to go through hundreds of movie list to find something that they may like. It was
difficult enough to look for list of movies on the Internet, moreover to browse through
the long list to find what they have been looking for. Therefore, it is essentially needed
for the user to use a recommendation system that can help them get through with this
process. As a result, the objectives for this project are to study the most suitable
algorithms used for a movie recommender system, to implement Collaborative
Filtering algorithm on a movie recommender system prototype and to test the
functionality of the movie recommender system, hi this project, a movie
recommendation system with the use of Collaborative Filtering is proposed to produce
a set list of movies for the users. This process will be done with the process of
calculation the similarity of the movies rated by the users and a recommendation list
will be produced through it. PHPMyAdmin is used to store the dataset and also acts
as a database for user information. The algorithm chosen was implemented using Java
Programming language and was tested using Root Means Square Error (RMSE)
formula.
5
CHAPTER 1
INTRODUCTION
This project is created to find a suitable algorithm that can be used for a movie
recommender system. Multiple researches have been done in order to find suitable
algorithms that can be used for a movie recommender system. Methods that are
recognized as approaches for movie recommendation systems including
Collaborative Filtering (CF), Content-Based Filtering (CBF), Hybrid Filtering
and Demographic Filtering.
1.1 Background of study
A recommender system is a system that makes predictions or filtration preferences
according to the user's choices. Using its users' likes and dislikes, a Recommender
System!RS) generates suggested items for users and searches things online in
rather quite a short tune (Katarya and Verma, 2017). There are various areas that
can implement the use of recommender systems such as movies, music, news,
book, research articles, search queries, social tags and products. According to Devi
and Parthasarathy (2018), Collaborative Filtering, Content-Based Filtering,
Hybrid Filtering and Demographic Filtering are the four approaches that mainly
are used in building a recommendation system. (Devi and Parthasarathy, 2018).
Collaborative Filtering has two common ways it can be implemented in winch are
User Based Collaborative Filtering and Item Based Collaborative Filtering (Devi
and Parthasarathy, 2018). User Based Collaborative Filtering takes the similar
taste between different users who rated the same items into consideration, then
based on the assumption, it recommends the items that have not been seen yet to
the active user. Meanwhile, Item Based Collaborative Filtering recommends the
items that have highest correlation.
