As you might already be aware, there is a disproportionate presence of "influencers" in forums and online platforms in China. In order to navigate our data gathering process to ensure we get the purest expression of real consumer beliefs, we use the aid of technology and human judgement. It is important to note that ultimately these are adjustable rules. We adjust these parameters to ensure the best quality of data is collected for each project.
First, any posts marked as "expert" is de-prioritized although there are exceptions to the rule. For example, the expert section of Zhihu, "zhuanlan" has non-professionals with a hobby interest in topics. Such "experts" do matter. So this is where human judgement does play an important role.
Second, posts where links to buy/discount codes etc. are provided are ignored. They bear a direct connection to paid influencers.
Third, in platforms where paid influencers receive a lot of engagement, the engagement from consumers is considered in the data, but the paid influencers original post is not.
Lastly, if we see patterns of the same users engaging repeatedly on a topic, those users are manually examined to ensure they aren't paid to engage, and if they are, their content is ignored.