Eamonn Keogh gave a tutorial at SIGKDD 2009 entitled "How to do good research, get it published in SIGKDD and get it cited". Someone forwarded the slides to me. I recommend them to you.
Particular ideas that I noted are:
- Have your paper written well in advance of the submission (this is necessary for the following two steps).
- Send a preview copy of your paper to authors whose work you are extending or refuting.
- Put your paper through a mock review process.
- Do your utmost to make your results reproducible. For instance, make datasets and high-quality code readily available.
- Provide auxiliary resources, such as tech reports that go into subsidiary material in more detail (fuller literature review, more exhaustive experimental results).
- For double-blind reviews, make supporting materials available anonymously online.
- Gather real data and make it available -- this can enormously boost your citations.
- Don't believe a result just because it has been published, and don't cite a paper just because someone else does (a number of telling and amusing examples are given in the slides). Reproduce the former, critically read the latter.
- Techniques with lots of parameters are suspicious. An amusing quote from John von Neumann: "With four parameters I can fit an elephant, and with five I can make him wiggle his trunk".
The tutorial is quite long, but informative and entertaining.