Photo by rawpixel on Unsplash
Research Question and Aim
The Yelp platform’s elite system has swayed our choices in many cases, however whether Yelp’s elite system is trustworthy is something that needs to be explored. Therefore, in this report, we analysed the users in the Yelp dataset who have reviewed/rated restaurants in Phoenix to find out if the ratings of elite users are close to those of the regular user, and if not, how to select the users who best represent the tastes of the regular user, i.e. those with the least difference in ratings from the regular user. At the same time, this paper tests whether machine learning algorithms can be used to predict users who truly reflect the preferences of the average user based on existing user information.
Conclusion
In general, there is an 8% chance that the elite users of the Yelp platform will mislead the regular user. It is possible to select a subset of users that are more representative of the majority of users’ tastes through some selection methods, i.e. the ratings of these users are more informative. And it is feasible to use machine learning algorithms to predict this subset of users based on their features.
If interested in the details of the report, please click on the PDF block at the top of the page.
Yu Cao
I am passionate about leveraging large language models for multimodal learning, with a specific focus on unsupervised domain adaptation and domain generalization.