Modeling

Reasoning

Implementation

Download and Installation

User Guide

Viewing Example Models

Viewing Example Profile

Exporting a Diagram

Reasoning

Label Propagation

Alternatives Selection

StarUML v1 User Guide

Publications and Adoption

Acknowledgements

Quantitative Reasoning

The RE-Tools toolkit uses a quantitative reasoning to recommend a desirable alternative for achieving a softgoal. The quantitative alternative recommendation is based on a weight-based selection, as described in this paper.

The following is an example of the weight-based alternatives selection.

In this example, user-defined weights for prioritized criteria (1.0 for high/!!, 0.5 for medium/!, and 0.2 for low) and for contribution links (1.0 for Make/++, 0.5 for Help/+, -1.0 for Break/- -, and -0.5 for Hurt/-) are displayed in gray color, and system-calculated scores in blue. The tool recommends the most desirable leaf-level alternative for Confidentiality using a depth-first selection algorithm. Specifically, the selection between Password and Biometrics is not made until the selection between Fingerprint and Retina Scan is made first, where the latter is selected for its higher score of 0.5, over 0.35 of the Fingerprint alternative. Biometric is then selected for its higher score of 0.5, over 0.15 of the Password alternative. A portion of Biometric’s score (0.3/0.5) is inherited from Retina Scan’s score that accounts for the positive and negative correlations with Trustworthiness and Cost (0.5 and -0.2 respectively). The tool labels the selected alternatives (Retina Scan and Biometrics) as satisficed (depicted by check marks) and uses the label evaluation procedure to qualitatively propagate the labels upward in the goal graph to determine the impacts on high-level goals.

Let’s look at the example above in more detail. Suppose we use the following weight assignments with values between 0.0 and 1.0 for different priorities (high, medium and low) and different contributions (Make/++, SomePlus/S+, Help/+, Hurt/-, SomeMinus/S-, and Break/--) where positive contributions receive positive weights and negative contributions receive negative weights to reflect the negative impacts.

Using the pre-defined weights, below shows the weights and resulting scores for the Retina scan option.

Notice the contribution link towards Biometrics is a solid line while the contribution links towards Trustworthiness and Cost are dashed lines. A solid line represents the contribution between a solution and its goal while a dashed line represents a correlation between a solution and one other goal.

Based on the weight assignment table, the following table shows the individual, cumulative score for all correlation links (against Trustworthiness and Cost), and the total cumulative score for the Retina scan option against the three selection criteria.

The individual score for each contribution is displayed over the contribution labels (0.2 for Make(Biometrics), 0.5 for Make(Trustworthiness) and -0.2 for Break(Cost) respectively. The final cumulative score of 0.5 is displayed over the Retina scan option in blue.

Comparing the cumulative score of 0.5 against 0.35 for the Fingerprint option, RE-Tools recommends the Retina scan option with a higher score by setting the value of its Label attribute to Satisficed, which is depicted by a check mark on the icon. Once it’s label is changed to Satisficed, the automatic Label Propagation Procedure is activated to evaluate its impact to label Biometrics as Satisficed.

Next step is to calculate the score for the Biometrics option for a comparison against the Password options. The total cumulative score is arrived from the following scores:

The reason the correlation scores for the Retina scan option are also included even Biometrics does not have any explicit correlation links with the two other criteria is that Biometrics represents the selected child option at the abstract option level. Otherwise, Biometrics would be an option that does not have any correlation with other criteria.

–

This page in Latvian (by Lucja Adamska)

Option	Criterion/Weight	Contribution/Weight	Score
Retina scan	Biometrics/0.2	Make/1.0	1.0 x 0.2 = 0.2
Retina scan	Trustworthiness/0.5	Cost/0.2	1.0 x 0.5 = 0.5
Retina scan	Cost/0.2	Break/-1.0	-1.0 x 0.2 = -0.2
		Cumulative Correlation Score	0.5 - 0.2 = 0.3
		Total Score	0.2 + 0.5 - 0.2 = 0.5

Option	Criterion/Weight	Contribution/Weight	Score
Biometrics	Authentication/0.2	Make/1.0	1.0 x 0.2 = 0.2
Retina scan	Trustworthiness/0.5	Make/1.0	1.0 x 0.5 = 0.5
Retina scan	Cost/0.2	Break/-1.0	-1.0 x 0.2 = -0.2
		Cumulative Correlation Score	0.5 - 0.2 = 0.3
		Total Score	0.2 + 0.5 - 0.2 = 0.5