Improving quality scores in RQA

Understanding the quality scores in RQA

IBM Engineering Requirements Quality Assistant (RQA) uses IBM Watson® Natural Language Processing (NLP) and a special requirements model to analyze requirements based on international industry standards. When you check requirements, RQA identifies issues based on those natural language and industry standards and returns a quality score. The score reflects the quality of the requirement: a lower score reflects lower quality and a higher score reflects higher quality. The scores help indicate which requirements need attention or improvement.

For DOORS® Next, the administrator creates the project-specific attributes that are named RQA Score and Issues Found by RQA that are used to view the quality scores and issues that RQA found in the list of DOORS Next artifacts. For more information, see Checking DOORS Next requirements with RQA customer-managed.

For DOORS, the RQA Score and Issues found by RQA are the default attributes that are created as a part of RQA for DOORS add-in. You can see the RQA results in the DOORS module by adding the RQA Score and the Issues found by RQA columns to the DOORS module.

RQA calculates a quality score for each requirement based on a starting value of 100 points. Points are deducted for each issue that are identified with the requirement as shown in the following table. You can also customize these values per your business to meet the needs of your company. For more information, see Customizing the RQA scoring model and Customizing the scoring model for RQA customer-managed for DOORS Next.

The individual deductions are added, and the total deductions are subtracted from 100. The minimum score for a requirement is 0, and the maximum score is 100. A requirement that does not have any issues receives a score of 100.

The quality scores provide reviewers with an objective basis for identifying requirements that need attention. For example, if you review a system with hundreds of requirements, you might choose to ignore requirements with scores more than 80, or focus immediately on requirements with scores less than 50.

Example 1

First, consider this requirement:

• The GPS System shall receive position data when in Power On mode.

The requirement does not have any of the identified issues and qualifies for a perfect score of 100.

Example 2

Then, consider this requirement:

• The GPS System shall receive position data when in Power® On mode and add the user position into all devices.

In this example, two requirements are combined into one. RQA detects a compound requirement and deducts 30 points.

Color coding

The quality scores are also color-coded. Scores less than 50 are red, scores in the range of 50 to 80 are yellow or amber, and scores more than 80 are green. The colors can help you identify which requirements need more attention.

"Look for" tip

For most issues that RQA identifies, the widget displays a Look for tip, which quickly identifies specific words or clauses that need attention. The Look for tip summarizes the logic that RQA uses. Look for tip might display "Look for: where there". "Where there" is an indicator of an escape clause, which is why RQA flag.

Show details

When RQA identifies an issue, the widget also displays a Show details link that you can click to expand for a description of the issue. For more detailed explanations and examples of the issue, see Improving quality scores in RQA.