Usually when we write papers we need to show some comparison between existing works and my work. Suppose by using existing approach we get 50% coverage but using my approach coverage is 70%. So my approach is better than existing approach. I have having some problem to define some criteria by which i can compare my approach with existing one. Can anyone help me out that how can i define these criteria? What should i consider while defining such criteria?
.
Unless your approach solves a problem which is completely new in the scientific litterature, you should find some pointers in the litterature proposing solutions to problems closely related to yours ; these pointers will suggest (possibly different) evaluation criteria
starting from this set of evaluation criteria, you can either criticize them and propose a new one on the ground of this criticism or choose one of them (or a few) to present a comparison of your approach with others'
.
You say "my approach is better than the existing approach". Why do you know this? How can you prove this?
@Ian - I think you need to use his previous sentence in conjunction with what you identified. In the previous statement he says if the existing approach covers 50% and his approach covers 70%, then the assumption holds. The question is what is being measured.
If, for example, we are talking compiler optimization, the resulting code might run 10% faster than the existing, but requires 20% more time to compile and takes up 10% more space on disk. Would 'my approach be better than the existing approach' be correct?
@Ian...I am saying "my approach is better than the existing approach" because existing approaches overlooked some criteria or some critical part which may cause error. On other word, existing approach doesn't consider those part which i have considered. I have generated some test cases which are not present in existing works.
@Jerry... I said "suppose by using existing approach we get 50% coverage but using my approach coverage is 70%".. this is just an assumption i am talking about. But problem is i cannot describe what to measure. I want to know that if i want to declare something like that what kind of measure should i consider. Is there any standard that i should follow?
@Muhammad - First, it always helps to identify your assumptions. Then explain/define 'why you are measuring' and 'how you are measuring', the 'what' will become more understandable. The best suggestion I can give is that you give details as to the 'why' and the 'how' - and then first apply them to your baseline (there's another ongoing discussion about baselines). Once you show the 'why' and 'how' applied to the baseline, you should be able to then do the same to your new method/implementation/idea to show what improves.
Usually you have some quality measures or requirements based on which you build your solution. You argue in your research that these requirements are important to consider for a good solution / approach. Based on these requirements you come up with an evaluation matrix or what you call criteria. Focus on a couple of these requirements and see how they were met compared to others. In other words, you are proposing a new solution to solve some issues in a better way. Therefore, you will compare based on these issues you considered in your approach. Additionally, based on your domain, you can consider other factors like complexity of the approach, costs ( time and resources), etc. I hope this helps!
However, depending on their formulations, the criteria underscore certain aspects of the set of analysed values. It is important to combine several complementary criteria for a comprehensive assessment. In addition, standard and common criteria should be compulsory to conform evaluation and comparison studies. With these considerations, the criteria were carefully chosen to correspond with various purposes.
- Badges
- Science topic
- Similar topics
- Human-Computer Interaction
- Collaboration