Question about Data Collection and Annotation Process for Research Use
Hello!
First, thank you for making this dataset available. It looks helpful.
We’re currently working on a university project focused on sentiment analysis using student feedback, and we’re considering using the ratemyprofessors-reviews-3-labels dataset as part of our research.
We had a couple of questions and would greatly appreciate any insights:
1- How was the data collected from RateMyProfessors? Was it collected randomly across instructors/universities or based on any specific criteria?
2- How was the annotation done for the sentiment labels (0 = Neutral, 1 = Negative, 2 = Positive)? Was this a manual process, an automated approach, or a mix of both?
This information would help us understand the context and reliability of the dataset better before including it in our analysis.
Thanks in advance
Hello!
1- The data was collected through webscraping across RateMyProfessors website by iterating through the number at the end of the URL for professors (e.g.: the 2233887 in https://www.ratemyprofessors.com/professor/2233887). Unfortunately, I did not note down the range I parsed through for this specific dataset, but from what I can see, the attributed number to each professor does not depend on their rating, department, or university. Thus, the dataset could be considered random to some extent.
2- The sentiment labels were done automatically by attributing ratings made by the students to labels. The process chosen was simple: Neutral == 3, Negative < 3, and Positive > 3. Albeit too simple now that I learned of possible decimal ratings.
If you want to tailor the dataset to your needs, I suggest taking a look at "ZephyrUtopia/ratemyprofessors_reviews". That dataset was collected through the same means as this one, although it parsed through a much wider range and contains more unfiltered information.
Good luck with your project!
Thanks a lot for the clarification and the helpful pointer. I really appreciate it!