BỘ GIÁO DỤC VÀ ĐÀO TẠO ĐẠI HỌC KINH TÉ THÀNH PHỐ HÒ CHÍ MINH BÁO CÁO TÓNG KẾT ĐÈ TÀI NGHIÊN cúu KHOA HỌC THAM GIA XÉT GIẢI THƯỞNG “NHÀ NGHIÊN cúu TRÉ ƯEH” NĂM 2024 Does AI Recommendation System Quality really matter in SNSs? A moderating role of eWOM on AI Recommendation System Quality and Attitude to Ho Chi Minh City’s Gen z Intention to Watch Short videos. Thuộc nhóm chuyên ngành: Kinh tể. Hồ Chí Minh, tháng 2/2024 Abstract Video platforms on social networks are increasingly exploding with the appearance of short videos, which enable users to access a large amount of information every day. In this context, the proposed system plays an important role in providing personalized experience for users, in a world with increasingly rich information.
Accuracy, novelty, and diversity elements in the proposed system have also made important contributions to the quality of the system, so as to increase the user experience and provide each user with short films suitable for his characteristics and preferences. In order to clarify the influence of recommendation system quality on consumers' attitudes and intention to watch short videos, this paper establishes and verifies the research model by using the theory of the Human Information Process and the S-O-R model. In addition, the study also regards eWOM as a moderating factor that affects the relationship between the quality of the recommendation system and short video watching intention, as well as the relationship between customers* attitudes and intentions. To analyze and verify the validity of the model, we surveyed 391 students and employees aged 15-26 in Ho Chi Minh City by online questionnaire.
The research model was re-verified by SPSS 20.0 tool and Smart PLS 4. Through investigation and experimental analysis, we proved the relationship between the above variables. Consequently, this paper will provide a theoretical basis for expanding the application scope of the S-O-R model, and at the same time provide a new perspective on the relationship between the suggestion system and the customer's intention under the adjustment of eWOM. Therefore, the problem of effectively applying an electronic suggestion and transmission system at the same time in short videos is put forward to marketers, which is attributed to providing novel, diverse, and accurate inclusive quality content.
Keywords: Recommendation System Quality, Recommendation System. Short video, Watch Intention, Customer Attitude, S-O-R, eWOM. ii Table of Contents Abstract. i Table of Contents.
ii List of Tables. iv List of Figures.iv List of Acronyms. Literature review and hypothesis development. Recommendation System Quality (RSQ).
Electronic Word-of-Mouth. Stimulus-Organism-Response (S-O-R). Human Information Processing. Recommendation System Quality, Recommendation Accuracy, Recommendation Novelty, and Recommendation Diversity.
Recommendation System Quality and Customer Attitude. Recommendation System Quality and the Intentionto Watch. Customer Attitude and Intention to Watch. The moderating role of Electronic Word-of-Mouth (eWOM).
Sample and data collection and procedure. Data analysis methods.1 Data Cleaning and Descriptive Statistics. Confirmatory Factor Analysis. Discriminant Validity Analysis.
PLS-SEM confirmatory factor analysis (CFA) results. PLS-SEM path coefficient analysis results. Conclusion and Implication. Discussion and Conclusion.
Limitation and Future Research. 47 iv List of Tables Table 1. Descriptive Statistics Table 2. Confirmatory Factor Analysis results Table 3.
Correlations between research constructs Table 4. Variance Inflation Factor Analysis results Table 5. Hypothesis testing results List of Figures Figure 1. Research Model Figure 2.
The platforms that are used to watch videos the most Figure 3. PLS analysis results for SEM V List of Acronyms AI Artificial Intelligence AVE Average Variance Extracted C. Composite Reliability CA Customer Attitude CFA Confirmatory Factor Analysis EFA Exploratory Factor Analysis eWOM Electronic Word-of-Mouth GenZ Generation z H Hypothesis R2 R square KMO Kaiser-Meyer-Olkin PLS-SEM Partial Least Squares Structural Equation Modeling ITT Intention to Watch RA Recommendation Accuracy RD Recommendation Diversity RN Recommendation Novelty RS Recommendation System RSQ Recommendation System Quality S-O-R Stimulus-Organism-Response model SNSs Social Networking Sites Sqrt(AVE) Square Root Coefficient TPB Theory of Planned Behavior VIF Variance Inflation Factor VND Vietnamese Dong 1 Chapter 1. Introduction AI has emerged and attracted academic attention since the 1980s with initial research on robots and expert systems (Chablo et al.
After nearly two decades, thanks to the growth of Big Data, the availability of computing power, and the development of AI techniques and technology-enabled tools have led AI to emerge as a trend in all fields (Bock el al., 2020; Overgoor et al. The success of AI in marketing practices has been proven through considerable research, especially from 2017 onwards. AI is widely applied in marketing to increase customer experience, personalization, automate tasks, and gain insights about consumers (Priyanga et al. In particular, user-centric AI systems have received special attention in developing intelligent systems that can flexibly respond to customer needs and preferences (Troussas et al.
User-centric AI systems are the ability to provide services suitable to the characteristics and preferences of each user, applied in creating a personalized experience for users. The development of artificial intelligence is in line with the changing expectations of customers, as they hope to showcase their individuality by having a unique product (Chandra et al. Marketers have already seen through this potential desire, and the concept of personalization has emerged with the rapid development of technology. The marketing department subsequently developed a considerable interest in this concept and conducted extensive related research (Polk et al., 2020, Lim el al.
Therefore, personalization is the use of information collected based on each customer’s social network purchases and interaction data to manage relevant interactions and provide customers with an excellent experience. Although achieving personalization is not easy, (Boundet et al., 2019) show that revenue has increased by 5% - 15% and single-channel marketing efficiency has improved by 10% - 30% when successfully applying this concept to marketing activities. In addition, in the context of the continuous development and expansion of the software industry, a large amount of data has been collected from past customers, users have higher expectations for personalized experiences and the potential benefits of providing personalized software solutions, and the demand for software systems that can meet the expectations and preferences of each user is also increasing (Sakarkar cl 2 al., 2021; Krouska et al., 2020; Pandey et al. In response to this, one of the most common examples of personalization in software is its use in recommendation systems (Troussas et al.
Academic research on recommender systems has matured substantially over the past two decades. This system is used in many different fields, such as online stores (Bhuvanya et al., 2023), social media platforms (Bekeneva et al., 2023), and video platforms to help users make decisions from a wide range of options. In this article, we will focus on the role of recommendation systems on social media video platforms. In addition, we will also focus on the key factor in providing users with a positive experience through recommendation systems, namely the recommendation system quality.
This is accuracy, novelty, and diversity (Rabab el al. With the continuous improvement and upgrading of technology, artificial intelligence is widely used in recommendation systems to analyze user video preferences, mainly through deep learning algorithms to establish user behavior models, predict their video needs, and provide accurate recommendations (Li, E. Since the launch of the short video platform TikTok in 2017, the platform has quickly gained popularity among global consumers, ushering in a short video craze. So far, the short video market is growing at an astonishing speed.
Short videos on social media platforms have gradually become the main way for most consumers to collect information (Song et al., 2021), and short video platforms such as TikTok, YouTube, Instagram Reels, and Facebook Reels have also become fiercely competitive in the marketing field. It can be seen that the development of science and technology in applying AI to recommendation systems along with the emergence of short videos on social networking platforms has provided users with a huge amount of information to reach. This leaves marketers with the challenge of achieving satisfactory performance of short-video marketing, and users with greater autonomy in proactively choosing what content they view or skip. Besides, the rapid expansion of social media has significantly changed customer purchasing behavior in the technology landscape (All et al.
They tend to seek information and suggestions from people who influence them or influential people on the internet. On social media platforms, both the quality and quantity of eWOM have a certain degree of influence on consumers' purchasing decisions. 3 In this context, our team conducted a research paper aimed at examining the impact of using a recommendation system on consumers' altitudes and intentions to choose to watch short videos, thereby evaluating the effectiveness in using recommendation systems to personalize the customer experience. There have been many research works on this topic (Li, E., 2023; Aggarwal el al., 2023) that have shown the relationship between recommendation systems in generating intentions in consumers.
However, previous studies only considered this relationship from a linear perspective and did not pay attention to external factors. Specifically, in the above context, how can electronic word-of-mouth factors moderate this relationship? This is a question that previous studies have not thoroughly addressed. For this reason, this paper aims to develop and supplement previous findings to address the following two research questions: 1. Does AI Recommendation System Quality affect Ho Chi Minh City's Gen z Intention to Watch Short videos? 2.
Does eWOM moderate the impact of Recommendation System Quality and Customers' Attitudes on their short videos Watching Intention? Based on the SOR model (Stimulus-Organism-Response) and Human Information Process Theory, the study proposes an integrated model to approach and analyze the level of influence and correlation between relationships. The survey was conducted through an online survey questionnaire, including 391 students and workers between the ages of 15 and 26 in Ho Chi Minh City, who have access to technology and grasp technology trends in the most timely manner. The research model is tested using statistical tools such as Cronbach's Alpha, exploratory factor analysis (CFA), structural difference analysis (Discriminant Analysis), Correlation Analysis, and linear structural model (SEM) using the bootstrap technique with replication of 5000 samples. In the next parts of the study, we distill the concepts and theories of the relevant documents of the previous system and present the methodological approach adopted in our study.
In Section 3, we provide the data collection and analysis methods. Section 4 presents and explains the obtained results. Finally, section 5 concludes the paper by summarizing the main contributions as well as discussing limitations and opportunities for further evaluation studies. Literature review and hypothesis development 2.
Recommendation System Quality (RSQ) Recommendation systems are important tools used by marketing departments to provide customers with product recommendations. Data scientists also use recommendation system analytics to evaluate the effectiveness of product and service recommendations (Ramjan & Sunkpho, 2023). These systems provide advice on products, information or services that may be of interest to the user. These are intelligent applications that assist users in the decision-making process when they want to choose an item from among a potentially overwhelming number of alternative products or services (Werthner et al.
Recommendation systems evaluation has evolved rapidly in recent years. A recommendation system is a key instrument in contemporary online market technologies, therefore it's important to look at the variables that can boost a user's chances of developing a fruitful, long-term connection with one. When recommendations accurately reflect a user's preferences, satisfy their needs, and satisfy their unique preferences, users are more likely to purchase from those recommendations. When recommendation quality is higher, users will likely perceive a higher level of usefulness and convenience as the system better meets their needs through improved accuracy, novelty, and diversity (Rabab All Abumalloh et al.
For offline evaluation, accuracy is the factor that standard for assessing the superiority of one method over another, with most research comparisons focused on tasks ranging from rating prediction to ranking metrics for top recommendation.