(Disclaimer: Quite subjective point of view, based on my experience more than any kind of reference or "authority". Ph.D student here, Computer Science, ML & NLP)
- Are titles written different for different areas of research? (i.e.
Computer Science, Maths, Neuroscience etc.)
I've no evidence about it, but I'm pretty sure it will be way different. Thus, my background might influence me a lot.
- Should one phrase it as a question or rather as a statement?
- Should one use humour and play on words?
I would say without hesitation, to me, humour has to be banned, and you should use a statement. You publish to give answer, not question. Question might be present at the end of your work to introduce what question does your work open, what are the next step. Still, I will be statement like "we may wonder what would...." (no "?" involved, you are wondering, not asking).
- Does one go with short and sweet?
- Or a bit longer and informative?
Here's the trap: it has to be short and informative. You have to tease future reader by telling him "I managed (or kind of) to do this, that way, it rocks right?".
This part will probably be the trickiest as you need to make people feel close to your research by showing what you work on, while not being too specific otherwise they will say "wow what's that?".
As a thumb rule, for a random paper (random = no hype = didn't read about it previously, no one told me to read it, I don't know/notice who wrote it), I will be interested if: the goal interests me or if the method interests me. Lets say if I'm interested about cars:
Making faster car or Using cars for food delivery.
Personaly what I see most is:
[WHAT I DID] <separator > [HOW DID I DO IT ], usually to present a solution to a general problem
- (1) is the high level goal like "Understanding human voice";
- (2) is "using", "with", ": ";
- (3) is the method, "Super Hugely Deep Neural Network" or approach "using semantic representations
[WHAT I USED] [WHAT FOR], using well known techniques for a context specific use or something not initially studied
- (1) Precisely name the technique
- (2) Describe your context/need.
[MY PRODUCT]: [DESCRIPTION], introduce a named product/innovation.
[WHAT I DID], when something is really new. Usually impressive work.
- e.g. "Efficient way of doing this" where the paper explains what this is, and how awesome it is.
Hope it helps, do not hesitate to comment, suggest edits etc. I find this discussion interesting and my really new to academic world, give me feedback!