Data licensing question

Question

So here's the situation: I built a database (text file) that consists of a few sets of data (for example, but not really), "car name", "car parts", etc.

The "car name" data and others are under no copyright, but for "car parts", I used two others sources of data, one licensed under Apache License 2.0, LGPLv3 and CC BY-SA 3.0, the other under GPLv2.

I then compared the two sets of data, one by one, manually, and selected certain "car parts" from one source, certain "car parts" from the other source and merged the two together (although I changed the formatting/layout), and sometimes I simply copied the data from one source intact and only changed the formatting/layout, and sometimes I simply edited the original data by removing or including certain other "car parts" myself (and, again, changed the formatting/layout). I also did a lot of fixing of errors, etc.

I then added additional sets of data in the other fields.

Now, my question is, because I manually built my custom dataset with the help of two open-sourced databases (which do not resemble my database any longer), do I absolutely must share this dataset publicly and make it available for all? The thing is, these particular "car parts" that I've selected and my custom formatting is something innovative and gives me an advantage over other people's projects who simply use one of the two open-sourced databases unchanged and in their original form.

I would like to make all this data publicly available and freely accessible to everyone via a website or app (free), but I don't want to make the dataset itself downloadable in its text form, because the goal of my project is to be innovative and offer unique new features, and I don't want some random competitive project to incorporate my hundreds of hours of research and data into theirs, essentially making my unique project just another "clone".

So what are my options, considering the mentioned licenses were never meant for (textual) data, but software and graphics/media, etc., and I am making my data freely available and accessible via a website/app for everyone - I just don't want to share the source database as a download?

Answer 1

The answer depends heavily on the context of you work and on potential users of your data.

First, there is no "must" about sharing your dataset. If the users want to get out more data than your website/app would allow them to do, they could always go to the original sources as they've been doing previously.

Secondly, if you decide to share your data (be it the whole database or just a part of it referring to a specific "car" with its "parts"), you must do it under the same licenses as the original sources do (or you might try to find one license, which satisfies them all, but it's not the easiest thing in the world). And as far as I can tell without more details, there might be already a potential problem with combining GPLv2 and LGPLv3.

Thirdly, it really depends on your area whether you should or should not distribute openly your work. In academia, for instance, it's highly recommended not only because we all stand for open data and global knowledge but because it will bring you citations, mentions, and scientific prestige. No fear that the others will use your work because you want them to use your work as widely as possible and to give you credit for it.
In private sector, however, it might be different, so it's up to you what to decide.

Finally, I should note that any data might be "scraped" from a website or an app. If somebody is really interested in getting a lot of data, what would stop them from doing it? You can try to build a protection against data scraping, of course, but at the end of the day, it might be more costly than sharing the data. Thus, you might want to add a way to contact you directly for big amount of data (say, "Here is my email, please send me an inquiry, describing your interest, your company/institution, and the intended purpose of data use").

Hope it helps.