Take the 2-minute tour ×
Academia Stack Exchange is a question and answer site for academics and those enrolled in higher education. It's 100% free, no registration required.

Some parts of our research fall out of our expertise. We thought about creating a Kaggle challenge just for this part. But we lack fund resources for the reward, so we could not offer any money. We are thinking now about offering co-authorship in a paper, do you think this would work? Otherwise, which other things would you offer as reward in Kaggle apart from money?

share|improve this question
1  
Have you thought about asking questions on related stack exchange sites to see if anybody has any clue about that problem? –  scaaahu Nov 7 '12 at 4:44
    
I will do that if in three days I do not get any answer here, but I had to start somewhere and I chose this nice place! –  flow Nov 7 '12 at 8:20
2  
You may well be able to offer other incentives (which is why I'm not submitting this as an answer), but coauthorship is tricky, since it is highly constrained: everyone who makes an important intellectual contribution must be a coauthor, and nobody else can be. For example, several other teams may submit ideas like the winner's (maybe with just a little less impressive write-ups or results); then they would all have to be offered coauthorship. In the other direction, if nobody's results are useful, then nobody can become a coauthor, even if they put a lot of time and effort into the contest. –  Anonymous Mathematician Nov 7 '12 at 13:35
    
@AnonymousMathematician I don't know how the authorship should work in this case but you are right that it's not something that can be simply traded. However, usually solutions very and when writing a paper I don't need to put in co-authors all people with similar approach - I need to acknowledge them, perhaps in form of citations. And IMHO it gives a natural framework when you choose your co-authors out of the solutions you want to use and only cite other solutions that make sense but you don't apply them directly. –  Piotr Migdal Nov 7 '12 at 15:50
1  
@AnonymousMathematician AFAIK solutions on Kaggle are not mathematical proof-style. There is a lot of one's own idea, a lot of technical fine-tuning; not just 0-1 or a few main ideas for the proof. Moreover, before actually asking participants, you see only a numerical result. So there is no risk of an unintentional plagiarism. And also one can set a threshold below which no authorship is proposed (personally I wouldn't promise any co-authorship but if the result accuracy is satisfying - asked the winner(s) for co-authorship). –  Piotr Migdal Nov 7 '12 at 23:53

3 Answers 3

It's true that fitting the Kaggle competition framework is a bit of a constraint, but if you understand that framework and see how your problem fits into it, I'd suggest you e-mail quote@kaggle.com with a description of the data & problem (and ideally sample data), even if you don't have funding. We can see what comes of it!

(Disclaimer: I work for Kaggle.)

share|improve this answer

This sounds to me like the textbook definition of a potential collaboration opportunity. I would speak with researchers in the computer science, machine learning, mathematics, and/or statistics departments and try to establish a long-term collaborative relationship. I would also speak with other faculty in your department to see how they handle this problem, which they likely have as well. It may be as simple as paying a grad or reasonably bright undergrad a few bucks an hour to crunch numbers. (Actually, you shouldn't take for granted that the grad student is reasonably bright... but I digress.)

This will be beneficial in both the short- and long-term, in a number of ways. In the short term, you'll hopefully get to analyze your data and generate a publication. In the long term, it's very likely that you'll need such expertise again, and you'll have the resources available for the analysis. Additionally, you'll find that having someone with analytics expertise on hand when devising whatever study you plan on doing can be immeasurably useful, as they will help you determine what measurements need to be taken and what data needs to be collected to ensure that you can get the most out of your dataset.

share|improve this answer

Some competitions have "Kudos" (or "Jobs", or "Knowledge") as a prize, see Kaggle competitions sorted from the smallest prizes. I don't how it does work with proposing a competition (maybe it is not free).

I guess if the problem itself fits in the Kaggle framework (which is a pretty strong constraint), and you can try it for free (I don't know if it is the case), then why not. Especially as even two "Kudos" competitions attract 50 and 153 teams, respectively. Just it is on you to make it as attractive as possible.

Of course, intellectual contribution requires a co-authorship. But if I had the right expertise, I would run for it :).

share|improve this answer

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

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