Abstracts need to include a brief summary of what sort of research you intend to propose. It is not good enough to simply say "I will also propose some new areas of research..." Academics that are reading your abstract will most likely already know what is happening in the field of research you are going to discuss, what they need to know is what special contribution your paper makes to that field.
Background information must always be summarized and kept as short as possible without losing its value. In a paper discussing the uses of game theory for AI decision making, you do not need two pages discussing game theory. You need to save the bulk of your paper for evaluating and discussing current work in the field. Remember, graduate level work is not about meeting a page limit or a word limit. It is all about sharing information at a high level. If you fill up forty pages of your thesis with background information, you are just going to be instructed to summarize it into three or four pages and that equals much lost time. So get it right the first time.
You need to use signposts to assist your reader in following your overall argument. Signposting can be accomplished by placing mini conclusions at the end of each section of your paper. Briefly summarize what you have discussed in that section and relate it back to your overall argument or idea.
It is not enough to simply review the work of authors in the field you want to do research in. You need to place those authors into discourse with one another. It is not enough to simply state that, "Simard also does work on Game Theory...". You need to explain why Simard's work is relevant to the field and then why the next author's work (let's name him Erdmann) you review is relevant to both Simard and the field. In what ways do Simard and Erdmann's research support each other? How do they contradict? Do they play off one another?
The comparisons are important because in your paper you are working towards establishing your own research proposal. Your proposal will rely on a strong understanding of how all of the previous research in the field is interrelated. In other words you are building up the reader's understanding of field specific research so that you can propose an idea that definitely advances understanding in a certain area of computing science. This goal is difficult to establish if your literature review reads like six or seven different authors and ideas that are unrelated.
It is also important to place authors into discourse with one another and discuss their ideas because you may make connections that others have missed or disagree with.
Finally, author review is also supposed to be a critical process not simply a summation. You need to engage with the author's work in a critical manner. Evaluate their methodology, their results, conclusions, and hypotheses. You need to use their shortcomings or successes to springboard your own proposals and research ideas.
Motivations for Research
It is often a good idea to discuss briefly your motivations for you research. This is something that can be done in the abstract of your paper or in your introduction. You need to let academics know why you feel that what you are doing is valuable. A motivation is not, "because I think video games are cool," but rather, "because research has shown that video games and games based design courses are effective teaching methods which could also be applied in..."
A conclusion is not just a paragraph that tells the reader that "I reviewed some literature in the area of research and made some new proposals." Rather a conclusion needs to sum up the most important aspects of your paper. In the case of the papers I reviewed the areas summed up in the conclusion should be as follows: what exactly you are proposing for future research, your motivation for proposing said research, and what value that research holds for the field.