Let me ask you a question, you’re in a new town and trying to work out where to eat tonight. On your phone you are scrolling through local restaurant reviews. Which reviews do you scroll past, and which do you stay and read?
In today’s post we are lucky to hear directly from one of our NUSC experts, a lecturer in Business Logistics and Transport at the University of Greenwich, Dr Duy Tan Nguyen. Dr Nguyen has been working at the university since 2024 after getting his PhD from HEC Montreal, the business school of the University of Montreal. This global academic previously studied in Vietnam and Australia. Famous for his work on data-driven and sustainable operations and supply chain management, today Dr Nguyen talks to us about his latest paper, ‘A framework for affinity-based personalized review recommendation’ published in Electronic Commerce Research with colleagues from Montreal, Warut Khern-am-nuai, Yossiri Adulyasak & Jean-François Cordeau.
“I noticed that the posts appearing or suggested in my Facebook newsfeed were mostly those liked by my friends and that the product reviews displayed to me on Amazon were from reviewers in my province. I guessed that this level of personalisation must have been informed by prior interactions between users on social networks and their locations. At that time, I could not find much literature analysing those variables. Luckily, the dataset I had access to included variables that could be used to measure or approximate those variables according to the literature. Therefore, I came up with that research idea and its conceptual model”
In this latest paper, Dr Nguyen creates a framework to identify and recommend reviews to online consumers based on the probability that they will like, comment, or re-read those reviews. Making use of advanced machine learning techniques, Dr Nguyen and colleagues demonstrate the potential of their work to enhance user engagement.
Dr Nguyen explains that the work behind this paper began in 2018, “around the time when Google search trends for Big Data and Industry 4.0 experienced a marked increase.” Dr Nguyen had been investing time in building his research and data skills, teaching himself R, Python and machine learning models.
I had completed a research project in which I interviewed a hard goods retailer and obtained interesting insights into how online stores could support and inform distribution network design. Therefore, I was then interested in applying the skills I was developing to research projects related to operations management informed by online data analysis. My professors and I were lucky to be granted access to suitable data for that research theme.
In his paper, Dr Nguyen identifies which variables online review platforms need to consider so that reviews appearing to consumers are relevant and personalised. “Whilst many factors, e.g., review age (or review recency) and review length, are popular determinants of review helpfulness or review relevancy, some others, e.g., reviewer-user similarity and locality, are unique to each reviewer-user pair, so using them for review recommendation can help personalise user experience.” His work helps to improve user experience.
User-review affinity indicates a reader’s positive attachment to a review. When a user reads a review and finds it relevant or useful, they may want to hit the like button, leave a comment or refer to that review later. These activities indicate that the user interacts more with the review and the platform, by continuing to use the platform or coming back later. Dr Nguyen used such interactions as a way to measure user-review affinity; by enhancing this, users see the most relevant content and organisations get a boost in customer satisfaction and retention. As Dr Nguyen explains:
Whilst reviewer-user similarity is an important factor in determining user-review affinity, other variables also exert a significant effect. As the products on the review platform under analysis are experience products, the opinions of those with similar interests to us are obviously relevant. Don’t we often ask friends and family for their advice before making several decisions? However, similarity is not the only determinant. A recent review (the review age is low) with a lot of helpfulness votes and from a reviewer with a huge following (social connectedness) is also likely to be recommended as many other users agree with the review or reviewer. Meanwhile, old information from a friend can be less likely relevant now.
Not only did Dr Nguyen find these factors to be important, he was also one of the first to empirically explore the role of locality – that is, the importance of the reviewers as local (geographically) to the person reading the review.
I also observed that when I moved to a different province, the location of the reviews on Amazon also changed automatically. I also noticed that my Facebook newsfeed depends on my previous interactions with friends. One of my professors also agreed with my observations and later published a paper on social interactions and peer evaluations.
What does this mean for business owners?
Customers are more likely to find relevant content and have positive interactions with the platform when product/service reviews are displayed according to their predicted user-review affinity, your business can boost customer satisfaction and retention. As previous interactions vary by reviewer-user pair, recommending reviews according to the framework proposed by my work can help your business personalise customer experiences and improve their convenience.
Ok, so we know that you may be more likely to take on board a review that is newer, rated as helpful, from someone with a big following, and from someone who is similar to you – but could there be a dark side to this?
The main purpose of the application is to benefit customers by helping them avoid scrolling through numerous reviews as most relevant reviews are recommended and displayed on the first page. However, from the recent Meta social media addiction lawsuit (BBC, 2026), I realise that reading reviews from our friends (followers/followees) on the review platform can keep us glued to the screen for too long as we are curious about other friends’ experience with the product/service we are interested in. This can be a dark side, which calls for further research.
Dr Nguyen conducted this research by looking at reviews of restaurants in Asia, so we asked- is this something we need to be mindful of? Can we expect similar results in other areas in the world?
Collectivist cultures tend to show an inclination to conformity. In other words, conformity to the norm is appreciated in collectivist cultures where members hold shared values within their group. This suggests that a review commonly liked by friends in your network is likely to exhibit shared values that you agree with. Research has also reported that collectivist customers rely more on reviews than their individualistic counterparts (see the cited research in the article). Indeed, my work shows that reviews with ratings deviating markedly from the majority have a negative effect, whereas research on Yelp in the USA, which scores high on individualism, reports such deviation exerts a positive effect (see the cited research in the article).
Beyond the paper, talking to Dr Nguyen offers fantastic insights into the process of research itself (including how long it takes to do good research) and important lessons learned.
The importance of the literature review
One key lesson Dr Nguyen highlighted in this project was the value of literature review in guiding research and making it work.
There are a lot of interesting lessons that I learned throughout this paper, from initial conceptualisation to publication. Even now, I still use these lessons to give advice to the students I supervise, for example, ‘begin with a literature review to identify the literature gaps that you are interested in addressing.’
Sometimes, I struggled with a step in this project. I spent a few days thinking about possible solutions, but in vain. Then, I decided to read some published research and eventually found some structured approaches. At that time, I wished I had read more carefully to avoid that struggle because it had already been addressed in the literature. I believe that the lesson has helped me develop a healthy habit of consulting the literature for practice and decision-making.
Dr Nguyen explains what this looks like in practice, “After having identified the overarching research theme and a few relevant variables, I began reading journal articles indexed on the Web of Science to obtain an overview of what had been studied in the literature, what variables had been examined and how they were measured. I spent a few weeks reading and summarising those papers, thanks to which I identified what variables were considered central to the research theme and what variables were considered interesting yet remained understudied.”
The importance of the theory
So, you have viewed the literature. You have identified the variables. You are good to go? Well, that depends. Do you have a theory? As Dr Nguyen explains, your findings do not say much without a theoretical framework to undersand them.
Still, my favourite thing about this research project was my realisation of the importance of theory. I began the project in 2018, around the time I started my PhD in Canada, so I was occupied with a lot of challenging coursework. I did write and revise research papers around that time, but my primary focus was on data analysis, and my use of theory was just to meet editors’ requirements. I did not clearly see its important role, nor did my professors explicitly clarify its importance for me. Later, in 2020, I read the articles of Miller et al. and Bansal et al., which first appeared online that year in the Journal of Supply Chain Management. Thanks to those articles, I understood that discussing the theoretical mechanisms underlying the observed relationships in the current sample can provide support for the generalisability and explicability of the results. Since then, I have been intrinsically motivated to look for theoretical underpinnings to inform hypothesis formulation in my research.
This focus on theory, and on the literature is exactly what advances science- creating new understandings of the world and progressing knowledge. Dr Nguyen explains further:
As several variables cannot be measured directly, my work builds on previous research to identify indicators that can operationalise those variables. Then, we use theories to inform our hypothesis formation, which is notably helpful when a variable of interest has received limited empirical research. Our work then can suggest relevant theoretical lenses and provide empirical research findings for future studies that explore the variable from a different perspective.
What’s next for you Dr Nguyen?
Dr Nguyen knows that this line of research is not yet finished and has several ideas for how to take it further, including exploring his findings in different contexts, countries and industries, and the potential to use experiments to test the robustness of the findings.
Would you like to get involved? Why not contact Dr Nguyen today!
Please join me in thanking Dr Nguyen for his time and insights into both the world of online reviews but more fundamentally on the process of research itself!
Discover more from Networks and Urban Systems Centre (NUSC)
Subscribe to get the latest posts sent to your email.
