Mining Meaningful Insights from Online Product Reviews with AI-Powered Analytics to Discover Customer Sentiment

Özge Cömert; Nurcan Yücel

Makale detayları

Mining Meaningful Insights from Online Product Reviews with AI-Powered Analytics to Discover Customer Sentiment

Yazarlar:

Özge Cömert | Nurcan Yücel

Yayıncı:

Fırat University

Ana Başlıklar:

Modelleme ve Simülasyon

Dergi:

Fırat Üniversitesi Mühendislik Bilimleri Dergisi Vol.35 Issue 2, pp.679-690

Belge Tipi:

Makale

Yayın Tarihi:

1/2023

Belge Dili:

Türkçe

Document ID (Org No.):

2001287560

pISSN:

1308-9072

Anahtar kelimeler:

Özet:

 E-Commerce is a digital commercial model in which the purchase and sale of products and services are realized through the internet. With the widespread use of the Internet and easier access to the Internet, e-commerce has an essential role in todays business world due to its advantages such as reaching a wider customer base, reducing costs and increasing transaction speed. Customer product reviews on e-commerce sites are an important marketing tool for companies. When managed correctly, positive product reviews can increase sales, while negative reviews can lead to the loss of potential customers. In addition, customer product reviews provide insight into product quality and customer service, and increase trustworthiness and customer loyalty. Therefore, e-commerce sites need to consider customer product reviews, manage them properly, and pay attention to feedback to increase customer satisfaction. This study explores customer sentiment by performing an artificial intelligence-supported analysis to extract meaningful information from online customer product reviews. In this context, a dataset containing a total of 2000 customer reviews published publicly on e-commerce sites was created within the scope of the research. In order to analyze customer sentiment on the created dataset, a new model based on the Long Short Term Memory LSTM model, one of the deep learning models, and text encoding was proposed. As a result of the analysis on the model developed with the supervised learning approach, 91.50 of accuracy was obtained. In other words, an artificial intelligence model that can predict customer sentiment with a very high accuracy rate and produce meaningful results from these reviews on product pages with hundreds or even thousands of customer reviews has been developed. This study indicates that businesses can benefit from artificial intelligence-supported algorithms to increase customer satisfaction, develop various product marketing strategies, manage online marketing processes and implement different business models.

 E-Commerce is a digital commercial model in which the purchase and sale of products and services are realized through the internet. With the widespread use of the Internet and easier access to the Internet, e-commerce has an essential role in todays business world due to its advantages such as reaching a wider customer base, reducing costs and increasing transaction speed. Customer product reviews on e-commerce sites are an important marketing tool for companies. When managed correctly, positive product reviews can increase sales, while negative reviews can lead to the loss of potential customers. In addition, customer product reviews provide insight into product quality and customer service, and increase trustworthiness and customer loyalty. Therefore, e-commerce sites need to consider customer product reviews, manage them properly, and pay attention to feedback to increase customer satisfaction. This study explores customer sentiment by performing an artificial intelligence-supported analysis to extract meaningful information from online customer product reviews. In this context, a dataset containing a total of 2000 customer reviews published publicly on e-commerce sites was created within the scope of the research. In order to analyze customer sentiment on the created dataset, a new model based on the Long Short Term Memory LSTM model, one of the deep learning models, and text encoding was proposed. As a result of the analysis on the model developed with the supervised learning approach, 91.50 of accuracy was obtained. In other words, an artificial intelligence model that can predict customer sentiment with a very high accuracy rate and produce meaningful results from these reviews on product pages with hundreds or even thousands of customer reviews has been developed. This study indicates that businesses can benefit from artificial intelligence-supported algorithms to increase customer satisfaction, develop various product marketing strategies, manage online marketing processes and implement different business models.

 E-Commerce is a digital commercial model in which the purchase and sale of products and services are realized through the internet. With the widespread use of the Internet and easier access to the Internet, e-commerce has an essential role in todays business world due to its advantages such as reaching a wider customer base, reducing costs and increasing transaction speed. Customer product reviews on e-commerce sites are an important marketing tool for companies. When managed correctly, positive product reviews can increase sales, while negative reviews can lead to the loss of potential customers. In addition, customer product reviews provide insight into product quality and customer service, and increase trustworthiness and customer loyalty. Therefore, e-commerce sites need to consider customer product reviews, manage them properly, and pay attention to feedback to increase customer satisfaction. This study explores customer sentiment by performing an artificial intelligence-supported analysis to extract meaningful information from online customer product reviews. In this context, a dataset containing a total of 2000 customer reviews published publicly on e-commerce sites was created within the scope of the research. In order to analyze customer sentiment on the created dataset, a new model based on the Long Short Term Memory LSTM model, one of the deep learning models, and text encoding was proposed. As a result of the analysis on the model developed with the supervised learning approach, 91.50 of accuracy was obtained. In other words, an artificial intelligence model that can predict customer sentiment with a very high accuracy rate and produce meaningful results from these reviews on product pages with hundreds or even thousands of customer reviews has been developed. This study indicates that businesses can benefit from artificial intelligence-supported algorithms to increase customer satisfaction, develop various product marketing strategies, manage online marketing processes and implement different business models.

 E-Ticaret, internet aracılığıyla ürün ve hizmetlerin alım-satım işlemlerinin gerçekleştirildiği bir dijital ticari model olarak bilinmektedir. İnternetin yaygınlaşması ve internete erişimin kolaylaşması ile e-ticaret, daha geniş bir müşteri kitlesine ulaşma, maliyetleri azaltma ve işlem hızını arttırma gibi avantajları nedeniyle günümüz iş dünyasında vazgeçilmez bir rol oynamaktadır. E-Ticaret sitelerindeki müşteri ürün incelemeleri, işletmeler açısından önemli bir pazarlama aracı olarak görülmektedir. Doğru yönetildiğinde, olumlu ürün incelemeleri satışları artırırken; olumsuz incelemeler ise potansiyel müşterilerin kaybına yol açabilmektedir. Bununla birlikte, müşteri ürün incelemeleri ürün kalitesi ve müşteri hizmeti hakkında fikir vermekle birlikte, güvenilirliği ve müşteri sadakatini arttırmaktadır. Bu nedenle, e-ticaret sitelerinin müşteri ürün incelemelerini dikkate almaları, doğru bir şekilde yönetmeleri ve müşteri memnuniyetini arttırmak için geri bildirimlere dikkat etmeleri gerekmektedir. Bu çalışmada, çevrimiçi müşteri ürün incelemelerinden anlamlı bilgiler elde etmek için yapay zekâ destekli bir analiz gerçekleştirilip, müşteri duyarlılığı keşfedilmektedir. Bu kapsamda, e-ticaret sitelerinde herkese açık olarak yayımlanmış olan ve toplamda 2000 adet müşteri incelemesi içeren bir veri seti araştırma kapsamında oluşturuldu. Oluşturulan veri seti üzerinde müşteri duyarlılığını analiz etmek üzere, derin öğrenme modellerinden biri olan Uzun Kısa Süreli Bellek LSTM modeli ve metin kodlamaya dayalı yeni bir model önerisi yapıldı. Denetimli öğrenme yaklaşımı ile geliştirilen model üzerinde yapılan analizler neticesinde 91,50 doğruluk elde edildi. Bir başka ifadeyle, müşteri duyarlılığını oldukça yüksek bir doğruluk oranı ile tahmin edebilen, yüzlerce hatta binlerce müşteri incelemesinin yer aldığı ürün sayfalarında, bu incelemelerden anlamlı sonuçlar üretebilen bir yapay zekâ modeli geliştirildi. Çalışmada, işletmelerin müşteri memnuniyetini arttırmak, çeşitli ürün pazarlama stratejileri geliştirmek, çevrimiçi pazarlama süreçlerini yönetmek ve farklı iş modelleri uygulamak üzere yapay zekâ destekli algoritmalardan faydalanılabileceğine işaret etmektedir.

Makale detayları

Mining Meaningful Insights from Online Product Reviews with AI-Powered Analytics to Discover Customer Sentiment

Yazarlar:

Özge Cömert | Nurcan Yücel

Yayıncı:

Fırat University

Ana Başlıklar:

Modelleme ve Simülasyon

Dergi:

Fırat Üniversitesi Mühendislik Bilimleri Dergisi Vol.35 Issue 2, pp.679-690

Belge Tipi:

Makale

Yayın Tarihi:

1/2023

Belge Dili:

Türkçe

Document ID (Org No.):

2001287560

pISSN:

1308-9072

E-commerce, sentiment analysis, machine learning, classificationE-ticaret

duyarlılık analizi

makine öğrenmesi

sınıflandırma

Sayı

Makale detayları

Mining Meaningful Insights from Online Product Reviews with AI-Powered Analytics to Discover Customer Sentiment

Yazarlar: Özge Cömert | Nurcan Yücel

Yayıncı: Fırat University

Ana Başlıklar: Modelleme ve Simülasyon

Dergi: Fırat Üniversitesi Mühendislik Bilimleri Dergisi Vol.35 Issue 2, pp.679-690

Belge Tipi: Makale

Yayın Tarihi: 1/2023

Belge Dili: Türkçe

Document ID (Org No.): 2001287560

pISSN: 1308-9072

E-commerce, sentiment analysis, machine learning, classificationE-ticaret

duyarlılık analizi

makine öğrenmesi

sınıflandırma

Sayı

Yazarlar:

Özge Cömert | Nurcan Yücel

Yayıncı:

Fırat University

Ana Başlıklar:

Modelleme ve Simülasyon

Dergi:

Fırat Üniversitesi Mühendislik Bilimleri Dergisi Vol.35 Issue 2, pp.679-690

Belge Tipi:

Makale

Yayın Tarihi:

1/2023

Belge Dili:

Türkçe

Document ID (Org No.):

2001287560

pISSN:

1308-9072