DEVELOPMENT OF E-COMMERCE WEBSITE RECOMMENDER SYSTEM USING COLLABORATIVE
FILTERING AND DEEP LEARNING TECHNIQUES
Medika Oga Laksana, Isma
Elan Maulani, Siti Munawaroh
Universitas
Muhammadiyah Cirebon, Indonesia
Email: [email protected],
[email protected], [email protected]
|
KEYWORDS recommender
system, collaborative filtering, deep learning, e-commerce website, performance,
user preferences |
ABSTRACT Recommender
system or recommendation system is becoming an increasingly important
technology on e-commerce websites to help users find products that suit their
preferences. However, the growing number of users and products makes finding
the right product difficult. Therefore, this study aims to develop a
recommender system on e-commerce websites using collaborative filtering and
deep learning techniques. Collaborative filtering is used to find
similarities between users based on their preferences, while deep learning is
used to improve the performance of the recommender system in generating more
accurate recommendations. The test method is carried out by comparing the
performance of the recommender system developed with the recommender system
that already exists on the e-commerce website. The results of the test show
that the recommender system developed is able to provide recommendations that
are more accurate and more in line with user preferences compared to the existing
recommender system. |
INTRODUCTION
In the era of
digitalization like now, the use of e-commerce websites is increasingly
widespread among the public because it makes it easier to buy the products
needed. However, the increasing number of users and products on e-commerce
websites makes the online shopping experience even more complicated. Users
often find it difficult to choose the right product, spend a lot of time
searching for relevant products and sometimes also find it difficult to
understand the product descriptions provided (Laksana, 2014).
To solve this
problem, the recommender system becomes a very important technology to help
users in finding products that match their preferences (Heryanto, 2018). In practice, the recommender system is used to
predict user preferences based on historical data, including information about
products that users have seen, purchased or rejected. Recommender system aims
to provide product recommendations that are relevant and in accordance with
user needs, so as to increase user satisfaction and speed up the purchase
process (Safitri, 2017).
However,
choosing the right technique in the development of a recommender system can be
a challenge in itself. Currently, there are several recommender system techniques
that have been developed, including collaborative filtering, content-based
filtering, and hybrid filtering (Prayoga & Kusnawi, 2022). Collaborative filtering is one of the most popular and
effective techniques in developing recommender systems. This technique works by
finding similarities between users based on their preferences, and then
recommending products preferred by users with similar preferences (Khusna, Delasano, & Saputra, 2021).
Although
collaborative filtering techniques have proven effective, there are still
drawbacks to using this technique (Wibowo & Munir, 2013). The main problem of collaborative filtering
techniques is the cold-start problem, where this technique cannot provide
recommendations to new users or new products that do not yet have sufficient
historical data. Therefore, more effective techniques are needed in developing
recommender systems on e-commerce websites (Wahyudi, 2017).
One of the
techniques that is currently developing in the development of recommender
systems is deep learning (Aisha, 2022). Deep learning is a machine learning technique that works by
utilizing highly complex neural networks, so as to predict user preferences
more accurately and effectively. Deep learning techniques have proven effective
in several applications such as facial recognition and voice recognition (Gusti, Nasrun, & Nugrahaeni, 2019).
In this study,
we developed a recommender system on e-commerce websites using collaborative
filtering and deep learning techniques (Hutabarat, 2022). The combination of these two techniques is expected to
increase the accuracy and effectiveness of the recommender system on e-commerce
websites. The test method is carried out by comparing the performance of the
recommender system developed with the recommender system that already exists on
the e-commerce website. It is hoped that the results of this study can
contribute to the development of a more accurate and effective recommender
system in the future (Nuryunita & Nurhadryani, 2013).
METHOD�� RESEARCH
This research uses quantitative research
methods with an experimental approach. Data collection is carried out through
the analysis of historical user data on e-commerce websites used as samples.
Collaborative filtering and deep learning methods are used in the development
of recommender systems. The test was carried out by comparing the performance
of the recommender system developed with the recommender system that already
exists on the e-commerce website. The resulting data were analyzed
using statistical methods to measure the accuracy and effectiveness of the
recommender system.
Population and Sample:
The population in this study were users of
e-commerce websites who had made purchases. Samples were selected using
purposive sampling techniques. The sample consists of e-commerce website users
who have made a purchase in 2021, with available and complete data.
Research Instruments:
The research instrument used is a data
collection technique from historical user data on e-commerce websites. The data
were analyzed using collaborative filtering and deep
learning algorithms to develop a recommender system. The performance of the
recommender system is measured using evaluation metrics such as precision,
recall, and F1-score.
Data Collection Procedure:
Historical user data on e-commerce websites
is collected and processed using collaborative filtering and deep learning
techniques. The resulting data is then used to develop a recommender system.
The test was carried out by comparing the performance of the recommender system
developed with the recommender system that already exists on the e-commerce
website. The resulting data were analyzed using
statistical methods to measure the accuracy and effectiveness of the
recommender system.
Data Analysis:
The data obtained were analyzed
using statistical methods to measure the accuracy and effectiveness of the
recommender system. The performance of the recommender system is measured using
evaluation metrics such as precision, recall, and F1-score. The test results
will be compared with the performance of the existing recommender system on the
e-commerce website. The analysis is performed using software such as Python, R
or MATLAB.
RESULT AND DISCUSSION
Result
The results showed that the development of a recommender system by
combining collaborative filtering and deep learning techniques can improve the
quality of recommendations on e-commerce websites. From the evaluation of the
performance of the recommendation system, it can be seen that the use of deep
learning techniques can significantly increase the accuracy and relevance of
product recommendations compared to using collaborative filtering techniques
alone. The results show that deep learning techniques can help models to learn
complex and abstract features in the product, so that the resulting
recommendations become more accurate and relevant (Firmansyah, Subroto, & Mulyono, 2022).
In conclusion, the development of a recommender system on an e-commerce
website using collaborative filtering and deep learning techniques can improve
the quality of product recommendations for users. Accurate and relevant
recommendations can help improve user experience and product sales on
e-commerce websites. Therefore, this recommendation system developed with
collaborative filtering and deep learning techniques can provide significant
benefits for e-commerce companies (Solichin & Painem, 2020).
Discussion
Recommender system is one of the technologies that is often used in
various e-commerce platforms to provide product recommendations to users. The
use of the recommender system aims to increase user convenience and attract
their interest in making product purchases (Pari & Kurniawan, 2021).
One technique that is often used in the development of recommender
systems is collaborative filtering. Collaborative filtering works by comparing
user preferences to find common ground between them and generating recommendations
based on those preferences. This technique is often used in movie or music
recommendation systems on platforms such as Netflix or Spotify (Kesuma & Iqbal, 2020).
In addition to collaborative filtering, deep learning techniques are also
increasingly popularly used in the development of recommender systems. Deep
learning allows models to learn patterns and features that are more complex and
cannot be learned by traditional methods. Deep learning techniques such as
neural networks and convolutional neural networks (CNN) can be used to generate
recommendations based on more complex data, such as images or text.
The development of recommender systems using collaborative filtering and
deep learning techniques can be done in several stages, including:
Data Processing
The first stage in the development of a recommender system is data
processing. At this stage, data from the e-commerce platform is captured and
processed to be able to be used in the model. Required data includes user data,
product data, and historical transaction data. Historical transaction data can
be used to train models and define user preferences.
Model Development
After the data is processed, the next stage is model development. At this
stage, collaborative filtering models and deep learning models are developed.
Collaborative filtering models can be developed using techniques such as
user-based collaborative filtering or item-based collaborative filtering. Deep
learning models can be developed using techniques such as neural networks or
CNN.
Model Evaluation
The next stage is the evaluation of the model. At this stage, the
developed model is evaluated using data that was not used at the time of
training. Model evaluation is performed using several metrics, such as
accuracy, precision, recall, and F1-score. These metrics are used to evaluate
how well the model can provide users with the right product recommendations.
Model Implementation
After the model is successfully evaluated, the next stage is the
implementation of the model. At this stage, the developed model is implemented
on the e-commerce platform and tested with actual users. At this stage, it is
also necessary to monitor to ensure that the model provides appropriate and
effective product recommendations for users.
The development of a recommender system using collaborative filtering and
deep learning techniques can improve the quality of product recommendations on
e-commerce platforms. Users will find it easier to find the desired product and
the e-commerce platform will become more attractive to users. In addition, by
utilizing advanced technology such as deep learning, the development of a
recommender system can produce more accurate and effective product
recommendations.
CONCLUSION
In today's
growing digital era, recommender systems are one of the most important technologies
in e-commerce platforms. The development of a recommender system using
collaborative filtering and deep learning techniques is the right solution to
improve the quality of product recommendations and attract users to make
purchases.
The collaborative
filtering technique works by comparing user preferences to find common ground
between them and generating recommendations based on those preferences. Whereas
deep learning techniques allow models to learn patterns and features that are
more complex and cannot be learned by traditional methods.
The
development of recommender systems using collaborative filtering and deep
learning techniques is carried out in several stages, including data
processing, model development, model evaluation, and model implementation. In
the evaluation phase, metrics such as accuracy, precision, recall, and F1-score
are used to evaluate how well the model can provide users with the right
product recommendations.
By developing
a recommender system using collaborative filtering and deep learning
techniques, users will find it easier to find the desired product and the
e-commerce platform will become more attractive to users. In addition, advanced
technologies such as deep learning can produce more accurate and effective product
recommendations.
In conclusion,
the development of a recommender system using collaborative filtering and deep
learning techniques is the right solution to improve the quality of product
recommendations on e-commerce platforms and can provide great benefits for
users and the e-commerce platform itself.
REFERENCES
Aisha, Dita. (2022). Sistem Rekomendasi
Toko Online Menggunakan Algoritma Collaborative Filtering Dan Content Based
Filtering. Universitas Islam Negeri Maulana Malik Ibrahim.
Firmansyah, Rusmal, Subroto, Imam Much Ibnu, & Mulyono, Sri. (2022). Sistem Rekomendasi Pada Pembelajaran Mobile Menggunakan Metode Cosine
Similarity Dan Collaborative Filtering. Transistor Elektro Dan Informatika,
4(1), 21�28.
Gusti, Ilham Gumantung, Nasrun, Muhammad, & Nugrahaeni, Ratna Astuti.
(2019). Rekomendasi Sistem Pemilihan Mobil
Menggunakan K-Nearest Neighbor (Knn) Collaborative Filtering. Tektrika-Jurnal
Penelitian Dan Pengembangan Telekomunikasi, Kendali, Komputer, Elektrik, Dan
Elektronika, 4(1), 26�32.
Heryanto, Bobby. (2018). Membangun
E-Commerce Dengan Menggunakan Metode Item-Based Collaborative Filtering.
Universitas Teknologi Yogyakarta.
Hutabarat, Kori Isabella. (2022). Implementasi
Algoritma K-Nearest Neighbors Pada Sistem Rekomendasi Buku Menggunakan Teknik
Collaborative Filtering. Universitas Medan Area.
Kesuma, Rahman Indra, & Iqbal, Amirul. (2020). Penerapan Content-Boosted Collaborative Filtering Untuk Meningkatkan
Kemampuan Sistem Rekomendasi Penyedia Jasa Acara Pernikahan. Jurnal Ilmiah
Fifo, 12(1).
Khusna, Arfiani Nur, Delasano, Krisvan Patra, & Saputra, Dimas
Chaerul Ekty. (2021). Penerapan User-Based
Collaborative Filtering Algorithm. Matrik: Jurnal Manajemen, Teknik
Informatika Dan Rekayasa Komputer, 20(2), 293�304.
Laksana, Eka Angga. (2014).
Collaborative Filtering Dan Aplikasinya. Jurnal Ilmiah Teknologi Infomasi
Terapan, 1(1).
Nuryunita, Kirana, & Nurhadryani, Yani. (2013). Pembuatan Modul Rekomendasi Pada Opencart Menggunakan Metode Item-Based
Collaborative Filtering. Jurnal Ilmu Komputer Dan Agri-Informatika, 2(1),
9�19.
Pari, Riki Aldi, & Kurniawan, Denni. (2021). Sistem Rekomendasi Keterampilan Dengan Metode User-Based Collaborative
Filtering Dan Tanimoto Coefficient Similarity. Respati, 16(2),
52�62.
Prayoga, Fajar Aji, & Kusnawi, Kusnawi. (2022). Smartphone Recommendation System Using Model-Based Collaborative
Filtering Method. Jurnal Teknik Informatika (Jutif), 3(6),
1613�1622.
Safitri, Marina. (2017). Rancang Bangun
Restful Web Service Pada Sistem Rekomendasi E-Commerce Berbasis Graf Neo4j
Dengan Metode Collaborative Filtering (Studi Kasus: Forbento). Tugas Akhir
Disusun Untuk Memenuhi Salah Satu Syarat Memperoleh Gelar Sarjana Komputer Pada
Departemen Sistem Informasi Fakultas Teknologi Informasi Institut Teknologi
Sepuluh Nopember, Surabaya.
Solichin, Achmad, & Painem, Painem. (2020). Sistem Rekomendasi Keterampilan Teknologi Informasi Dengan Metode
User-Based Collaborative Filtering Dan Log-Likelihood Similarity. Cogito Smart
Journal, 6(2), 141�154.
Wahyudi, Indah Survyana. (2017).
Mesin Rekomendasi Film Menggunakan Metode Kemiripan Genre Berbasis
Collaborative Filtering. Institut Teknologi Sepuluh Nopember.
Wibowo, David Eko, & Munir, Rinaldi. (2013). Sistem Rekomendasi Jual Beli Barang Dengan Memanfaatkan Metode
Collaborative Filtering Dan Basis Data Graf (Studi Kasus: Bukalapak. Com). Tugas
Akhir, Sch. Of Elect. Eng. And Inf., Inst. Technology Of Bandung.
Copyright holders:
Medika Oga Laksana, Isma Elan Maulani, Siti Munawaroh (2023)
First publication right:
Devotion - Journal of Research and Community Service
This article is licensed under a Creative Commons Attribution-ShareAlike 4.0 International