An online reviews-driven large-scale group decision-making approach for evaluating user preference of fresh fruit on e-commerce platform

In the digital marketplace, a vast ocean of text flows from millions of shoppers, each leaving behind a trail of words about what they bought, what they loved, and what disappointed them. For sellers of fresh produce, these words are a lifeline, yet they are also a source of confusion. Unlike a toaster or a pair of shoes, fresh fruit carries an inherent uncertainty; a fruit that looks perfect on the screen might arrive bruised, unripe, or simply lacking in flavor. This unpredictability makes it difficult for online stores to know exactly what their customers want. To navigate this, researchers have turned to a field of study known as large-scale group decision-making. This approach treats the collective voice of thousands of shoppers as a single, massive group trying to agree on a solution. Instead of asking a few experts to guess what the public wants, this method listens to the crowd, organizing their chaotic opinions into a clear, shared understanding. It is a way of turning the noise of individual complaints and praises into a signal that can guide better business choices.

Building on this foundation, a team of researchers from Northwest A&F University in China has developed a new way to listen to this crowd, specifically for the world of fresh fruit on e-commerce platforms. They focused on the massive amounts of online reviews generated by shoppers on a major Chinese shopping site. Their goal was to move beyond simple star ratings and dig into the actual words people used to describe their experiences. By treating every product review as a unique voice in a giant conversation, they created a system that could extract the true preferences of the shoppers. They found that when people buy fresh fruit online, they are not just looking for the fruit itself; they are deeply concerned with the service surrounding it. The researchers discovered that customers care most about how the seller handles problems, the reliability of the platform itself, and the condition of the store. These factors often outweigh the specific taste or size of the fruit in the minds of the disappointed or satisfied buyer.

To make sense of the millions of reviews, the researchers first had to clean the data, stripping away empty comments and meaningless characters to leave only the genuine feedback. They then used a computer program to identify the most common words shoppers used, such as "customer service," "delivery," "fresh," and "taste." These words became the building blocks for a new kind of map. Instead of treating every shopper as an isolated individual, the researchers looked for patterns in how people's opinions overlapped. They grouped the shoppers into smaller communities based on how similar their reviews were. Imagine a large room where everyone is talking; this method helps identify the smaller circles of friends who are discussing the same specific issues, allowing the researchers to see the different perspectives within the larger crowd. They used a sophisticated clustering technique to organize these groups, ensuring that even small, quiet groups of shoppers with unique concerns were not lost in the noise of the majority.

Once the shoppers were organized into these communities, the researchers faced the challenge of getting everyone to agree. In any large group, there are always some voices that disagree with the rest. The team developed a strategy to find these dissenting voices and gently guide them toward a shared view without forcing them to change their minds too drastically. They calculated the cost of asking someone to change their opinion, considering how influential that person was within their community. If a person had many connections or a strong voice, changing their mind required more effort. The system used an optimization method to find the path of least resistance, adjusting the opinions of the fewest number of people necessary to reach a point where the entire group felt satisfied. This process ensured that the final result was a true consensus, reflecting the collective will of the shoppers rather than just the loudest voices.

With the group's opinions aligned, the researchers then had to decide which preferences mattered most. They applied a mathematical framework that treated the shoppers' opinions as fuzzy, or slightly uncertain, information, which is a more realistic way to handle human feelings than strict numbers. By analyzing the relationships between different preferences, they assigned a weight to each one, determining its importance in the final decision. The results were striking. The analysis revealed that "customer service" was the single most important factor for shoppers. This was followed closely by the reputation of the platform, specifically the shopping site itself, and the quality of the store. While the taste and freshness of the fruit were important, they ranked lower than the trust and support provided by the seller and the platform. This suggests that when buying fresh food online, the safety net of good service and a reliable platform is what gives customers the confidence to make a purchase.

The researchers did not stop at theory; they built a working tool to put these findings into practice. They created an intelligent decision-support system that can be used by e-commerce platforms to answer customer questions and provide personalized recommendations. This system can read the reviews, understand what the customers are really saying, and offer advice based on the collective wisdom of the group. For merchants and platforms, the implications are clear. To sell more fresh fruit, they must focus on their after-sales service, ensure their logistics are fast and reliable, and maintain high standards for their stores. The study shows that while the fruit itself must be good, the experience of buying it—how it is delivered, how problems are solved, and how trustworthy the seller appears—is what truly drives satisfaction and repeat business.

The researchers also tested their method against other ways of analyzing data to ensure it was robust. They compared their approach with traditional methods and found that their system produced more stable and reliable results. They checked their work by changing the settings of their models to see if the results would shift, and they found that the ranking of the most important preferences remained consistent. This gives confidence that the findings are not just a fluke of the specific data they used, but a genuine reflection of how shoppers think. The study concludes that by listening to the crowd and using smart tools to organize their voices, online sellers can better understand their customers. This leads to a better shopping experience, fewer complaints, and a more successful market for fresh produce. The work highlights that in the digital age, the key to selling perishable goods is not just in the quality of the product, but in the quality of the conversation between the buyer and the seller.