Project
Provincial Undergraduate Innovation Training Program
My role
Model construction and report writing
By the midpoint
More than 10,000 Taobao product samples compiled
What initially attracted me was how real the problem is
Our undergraduate research project focuses on the relationship between regional public-brand rice, e-commerce platforms, and traceability systems. Put simply, when an agricultural brand makes information such as origin, quality inspection, and traceability more transparent, will consumers actually be more willing to buy? Will its sales, pricing, and brand perception genuinely change?
I was drawn to this topic because it is not a purely conceptual question. It connects platforms, data, consumer behavior, and brand trust, and it is directly tied to real-world agricultural branding and digital transformation. It is complex, but not in an abstract way. It is a kind of complexity that can be gradually unpacked.
This was the first time I clearly realized that research does not begin with a polished conclusion. It begins with asking the right question, and then building it step by step.
What was truly difficult was not the idea, but constructing the data
The most memorable part of this project was the sheer workload in the data stage. We did not start from a clean, ready-made dataset. Instead, we began from e-commerce platforms, built our own scraping logic, defined variable systems, iteratively expanded keywords, and then went through multiple rounds of sampling, supplementing, cleaning, and verification.
By the midpoint, we had compiled over 10,000 product samples from Taobao and built a review corpus. Many of the most time-consuming tasks were not coding itself, but the details that required manual validation. For example, how to standardize product specifications, how to classify brand names, and whether “traceability information” embedded in product descriptions, without standardized fields, should count as valid disclosure. These could not be determined purely by automated methods.
Over time, I realized that the most demanding part of research is not “not knowing how to build models,” but transforming messy, fragmented, and unstructured information into something that can be meaningfully studied. This process is slow and tedious, but it is fundamental, because the credibility of all subsequent analysis depends on it.
This project was also the first time I truly understood what a “research framework” is
Previously, I saw a framework as something abstract, perhaps a diagram or an outline. But after completing this project, I see it as the ability to progressively narrow down a real-world problem. You need to identify core variables, determine data sources, decide which factors to control for, and be clear about the exact question you aim to answer.
In this project, I was mainly responsible for model construction and report writing. For me, the most valuable training was not about using sophisticated terminology, but about explaining a problem clearly and demonstrating why the research holds. The strength of a study often lies precisely there.
If I had to summarize my biggest takeaway, it would be this: it was the first time I seriously connected structure and expression. Instead of chasing results alone, I learned to align the problem, method, and conclusion into a coherent line. That shift has influenced how I approach many things afterward.