Analysis scheme for SMM tools

This scheme is based on the so called "feature analysis", described in "DESMET: A method for evaluatiing Software Engineering methods and tools" in 1996 by Barbara Kitchenham. This will be a modified version of this feature analysis, in which the following steps will be performed:

1. categorization of the tools
2. benchmark of the single properties using the benchmark scheme
3. mark if the use of a tools API is for free
4. display the results in graphical way
5. compute the point for every tool using a formula
6. rank the tools

The categorization of the tools is important because every property will be priorized within a category. This will be done, because it is one way to manage the different needs concerning the tools. In order to priorize every property will get a value between 1 and 4, with 1 being "not important" and 4 being "very important". There will be a difference between simple consumers of the tools and such companies who willing to intergrate the into own solutions. For them the interesting categories are the same but the focus on properties is a different.

This is a list of the properties, which will be considered during the analysis:

1. API
2. data analysis (quantitative)
3. data analysis (qualitative)
4. sentiment analysis
5. data extraction
6. engagement possibilities
7. ways of personalization
8. recommendations concerning the reaction of the results
9. language support
10. price-performance ratio
11. support of different sources
12. usability

Therefore the benchmark of these properties will use a system of 4 levels, starting with 0 going to 3:

• 0 points: property not fullfilled
• 1 point: property insufficiently fullfilled
• 2 points: propery sufficiently fullfilled
• 3 points: propery very sufficiently fullfilled

The detailed benchmark scheme won't be presented here, but will be in the final thesis. The goal is to achieve objectivity, which won't be possible in this work, because the only person testing all these tools will be the author. So based on the final work one thing to do, would be to implement the analysis as web application and use the principle of "Wisdom of the Crowds" in order to achieve the objectivity.

Last but not least, there will be the computation schema for the point. In this case there are two possible scenarios - decisson still open:

1. Every property has a value between 0 and 3 by doing the benchmark. These values will be multiplied with the priority of the property concerning the category the analysed tool is in. The sum of theses products lead to the points of the tool. The maximum of points will be 144, by assuming that every property was priorized with 4 and got the value 3)



fomula - feasible for both scenarios

2. The second scenario would be, to depend the maximum of points to a certain category. In this case to compute the maximum of points it would be assumed, that a tool gets the value 3 concerning every property. But now it will be multiplied with the real priority choosen for the property in the category. So the maximum of points will differ in each category. So to make a comparison feasible, not the points itself should be compare but the percantage of the reached points.