8

I have a website that depends on reviews. In the initial stage I will have a classic chicken-egg problem. I need reviews to attract readers, and I need readers to attract writers.

After some research I found two open spreadsheets crowdsourced with anonymous reviews. This is gold for me, I can import some of these reviews to my platform (the good ones).

All reviews are completely anonymous in these spreadsheets. I don't see any licence or anything related. Apparently the people are just filling forms using google forms from some facebook groups.

Can I use this information in my app? Should I say that explicitly associating all content to the same user?

E.g. All reviews associated to the user crowd_from_web or should I simulate multiple users posting reviews from these spreadsheets?

  • Why don't you flush this question out a little more and tell us what kind of comment site. There are lots of review sites, but that doesn't mean you don't have a great idea that people wouldn't come and help you build if they believe in it too! For example: You're an expert on Yak herding. I would guess there is not a site to help people read about those strategies and tools and help them pick the best ones, you'd probably have a money maker (and a community of like-minded friends). But on the other hand, lets say you want to review cars - - probably no way you can get traction. – Baronz Apr 29 '17 at 20:34
5
+50

Here are some factors you should consider:

License: You said you didn't see a license, but the absence of a license doesn't mean it's free to use. It means you need to look harder. (Reviews are probably substantial enough to be covered by copyright law in most countries, too.) Try to find the form, or the Facebook users who circulated it, or something. (Try Googling for the text of one of the reviews to see where they've been published.)

If (hypothetically) Yelp harvested reviews from UrbanSpoon, it's a good bet that the latter would complain. So the first question you need to answer is whether you can use the data at all.

Assuming you can...

Data model: Will your web site allow a user to post multiple reviews of the same item? If not, then if you were to use a single aggregate user like crowd_from_web, you would have to build special cases into your code. Special-case code can be a sign that you're doing something that's not quite right in some fashion.

Aggregation only, or deeper data-mining? Could the plans for your web site ever include any sort of "users who liked X also liked Y" logic? If so, you'd be at least throwing valuable data away, and possibly messing up your data set besides, if you lump all the imported reviews together under one user. If the data has any way to tell users apart -- user numbers, for example -- then you're better off retaining it. It's easy to present it as a single source later if you decide you don't care about this at all, but you'll never be able to go back.

If the reviews are completely anonymous without even user numbers or the like, you're probably better off simulating one user per review instead of aggregating them. The user names you create for these imported quasi-users can reflect the source in some systematic way, and then you can give proper credit to your source in an "about" page or similar. ("Reviews posted by users with names of the form user_### were imported from such-and-such source.") Then use a slightly different pattern for newly-created users.

  • Thank you. I am thinking about create some users, trying to simulate real activity using the rules of platform. Saying explicitly that a bunch of users are fake (user_###1, user_###2, ...) could lead to a sentiment of "everything is fake", I think. – anvd Mar 31 '17 at 19:32
  • 1
    @anvd I'm sorry your bounty didn't draw more answers. I hope mine was helpful anyway. Good luck with your new community! – Monica Cellio Apr 3 '17 at 23:01

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.