The title should be short and say what the paper contributes. One common mistake is to put the word "novel" in the title, as in "A Novel {insert overdone type of application here}". The word "novel" should be implicit in every research paper title and never need writing.
The ACM has a text format for papers published in its conferences [ACM Text Format]. This format applies well to research conference publications in general, and is used here as an example for discussion.
Given the ACM format, full papers are typically 8-10 pages. Some rules of thumb to follow, loosely, for how large to make the different sections are:
The abstract should have no references or expanded acronyms. It should focus on what is novel about the paper: what it contributes.
In the introduction, it is handy to mention the major concepts and tools used, give their references, and expand their acronyms. The last paragraph of the introduction should lay out the structure of the rest of the paper.
In many cases, a background section is not necessary. It can often be better having its contents distributed throughout the main body sections, placed right where the become applicable. If the background section and introduction together go beyond the second page, it is a sign that too much background information is given. The authors should assume that the reader, or review, has chosen to read the paper for a reason. That is, most readers would not have chosen to read the paper if they weren't already familiar with the field and somewhat convinced of the merits of the paper's topic. Allow cited references to be places the reader can go for more information, rather than putting that information in the paper when most readers will not need or want it.
Content can sometimes be moved between the main body sections to change the number of sections or to equalize their length. This guideline should only be followed, of course, if it improves the readability of the paper. Often it does, by dividing the paper up into processible, equally-sized pieces. But it only works it the content is arranged and reworded properly. If no way can be found to do so, then unequally long or too many sections is a better alternative.
The summary or conclusion should not contain any new information. It should also not be a rewording of the abstract or introduction. Its wording should account for the fact that the reader has just read the paper, whereas the abstract and introduction assume the reader is not familiar at all with the paper's contents. Thus, the summary is reminding the reader of the keypoints and putting them in context with one another while they are fresh in the reader's head. A conclusion, similarly, takes these key points while they are fresh in the reader's head, and with them guides the reader to understand the paper's main points.
Redundant references should be avoided. Some reference sections read like gratuitous name-dropping, and detract from the readability of the section and the citations to them. There should be one reference per topic. Each reference should uniquely describe and address its topic. References can effectively be inherited, since they are intended in part to be places the reader can go to for more information. Thus, let the reader go to them for more information, such as additional references on the topic.
The authors should make clear kn the beginning what type of paper they are writing. Two types of papers to distinquish between are ones that introduce new research ideas and ones that are case studies. If a paper is a case study, the authors should focus on the techniques used as having been established, and then comment on how the use of them for the given application compares with that of other applications. If, on the other hand, it introduces new research ideas, these should be focussed on from the beginning. Only components of the implementation that relate to these ideas should be discussed.
When submitting a paper, write it in the format for the conference. This makes is easier for the reviews to compare with other papers. It also makes a positive impact on the reviews that the authors took the time and consideration to use the conference format. And in the end, it makes for less work if the paper is accepted and then the conference format must be used anyway.