Collect App Reviews - Githubissues

christina-tkach commented 2 years ago

The goal of this issue is to collect a minimum of 1000 reviews for any 10 cryptocurrency-related applications (100 each). These could be the apps of the blockchain wallets, crypto custodians, or any crypto projects.

Collect reviews for the apps of your choice and make sure that you define where the reviews are coming from - Apple App Store or Google Play Store.
Identify sentiment score for each review using one of the existing sentiment analysis tools.
Identify either geolocation, or language for each review.
For the final deliverable, create:

CSV-file with the following structure: app name username timestamp app review text sentiment score country (language) marketplace*

*marketplace - Apple App Store or Google Play Store

short report with graphs (based on the reviews that you collected) and basic descriptive statistics

Additional Resources

Please email challenge-submission@blockshop.org with your solution, and don't forget to include a link to this issue and attach your resume. Don't hesitate to ask us questions by commenting in this issue or emailing us at challenge-program@blockshop.org

mariiatarasova commented 2 years ago

I'm done

ChandlerProvence commented 2 years ago

Good evening, I have submitted the requested CSV to the provided email. Thank you for your time and consideration.

KY-39 commented 2 years ago

I'm done, nice task.😏

nemaziz commented 2 years ago

I'm done! Thanks

MarkAvilin1 commented 2 years ago

I'm done, thanks

geoburdin commented 2 years ago

Should we try to maintain a representative distribution of the number of reviews by country? On a sample of one hundred reviews per application, the selection of these reviews becomes too sensitive to outliers if you look at the non-major country sub-sample As an example: 70 out of 100 reviews could be in English, but only 5 in German, which does not reflect any patterns in the evaluation of the app by German-speaking users because of the small sample size.

This problem of undersampling can be conditionally solved by expansion of the dataset and collecting a lot more reviews, but I'm not sure if this is required and implied in the task, if geographic distribution is not the feature we would like to extract

tanya-ling commented 2 years ago

Hi! It seems a little strange to me that we are not supposed to include rating (the number of stars in the original review) in the data. Is that correct? Should I just remove this information from the data?

LastShekel commented 2 years ago

somacruz@bk.ru Done

christina-tkach commented 2 years ago

@geoburdin I agree, but the main objective here is not to collect a non-biased geo distributed sample, so that's alright to stick to the numbers mentioned in the description of the issue. That's also ok to collect a bigger sample if that fits your own goals when solving the task.

christina-tkach commented 2 years ago

@tanya-ling that's correct, you can stick to the fields mentioned in the description of the issue