RQA SaaS architecture

The IBM Engineering Requirements Quality Assistant (RQA) service provides a widget where users can check the quality of their requirements with a simple click.

In IBM Engineering Requirements Management DOORS® Next (DOORS Next), the RQA widget is accessible from the mini dashboard. It works in its own IFrame and interacts with DOORS Next through an API.

On IBM Engineering Requirements Management DOORS (DOORS) systems, you install RQA, log in, and then analyze requirements in a window that you access from the Requirements Quality menu.

RQA uses the IBM Watson® Natural Language Processing (NLP) to analyze requirements.

RQA analysis workflow

The following diagram shows the architecture and the workflow of using RQA to analyze requirements. A DOORS 9 user is shown as customer X and a DOORS Next user is shown as customer Y. Green Arrows show that customers get the client from the cloud. For DOORS Next, the widget is hosted in the cloud. For DOORS, the installer is hosted in the cloud.
Requirements analysis workflow with RQA
To analyze requirements, follow these steps:
  1. Log in to RQA by using App ID authentication (blue arrows).
  2. In DOORS Next or DOORS, select the requirements to analyze. RQA sends the requirement text to the IBM Watson Natural Language Processing (NLP) (yellow arrows).
  3. Receive the requirements analysis results from RQA, including these parts (yellow arrows):
    • Quality score
    • List of issues
    • Solutions to improve the requirement text
  4. In DOORS Next, you can run Jazz® Reporting Service reports to summarize the RQA insight on the quality of the requirements.
  5. You can improve the requirements by using the advice from RQA and recheck the updated requirements as needed.
  6. You can also use Teach Watson™ to enhance the analysis model for your situation by sharing feedback on the requirements guidance that RQA provides.

RQA components

IBM Watson Natural Language Processing (NLP) and IBM Watson Knowledge Studio are important building blocks of the RQA architecture.
  • IBM Watson Natural Language Processing (NLP)

    IBM Watson Natural Language Processing (NLP) helps you detect potential ambiguities and generates real-time scores to assess the quality of requirements.

    For RQA, a unique model overlays the core natural-language processing capabilities. The model contains requirement guidelines that are aimed to bring consistency according to international industry standards, such as those from International Council on Systems Engineering (INCOSE) and NASA. The model can be updated through IBM Watson Knowledge Studio.

  • IBM Watson Knowledge Studio

    IBM Watson Knowledge Studio is a cognitive learning tool that provides meaningful insights from data. Knowledge Studio analyzes requirement statements, identifies relationships and trends from the data, and creates rules and guidelines to complement the ones that are programmed into it. With practice, the tool can analyze faster and more comprehensively than human analysts.

    By default, no runtime data or requirement text is stored. You can enable the Teach Watson feature to train and improve the capabilities of the scoring model. If you enable Teach Watson, requirement text and customer feedback are stored.

Security considerations

As a cloud-hosted service RQA embeds security as an important aspect of its architecture. Your organization's data is protected, focusing on the following areas:
  • Compliance: External standards, which set benchmarks for security.
  • Authentication: APP ID that is used for RQA authentication.
  • Authorization: Assuring that users, devices, and applications have permission to access your organization's information.
  • Encryption: Data in transit is encrypted over HTTPS by using TLS 1.2 and it's only readable by authorized parties. The data is not persisted.
The requirements are processed by the RQA service without being stored in any external repository or Watson NLP. There is no data at rest outside of DOORS Next or DOORS.
  • The RQA data centers are hosted in Germany.
  • App ID, hosted in Germany, is used for RQA authentication and stores user IDs.

Administrators can enable or disable the Teach Watson feature. See enabling or disabling Teach Watson. When the Teach Watson feature is enabled, users can provide feedback on the RQA requirements guidance or if they disagree with an issue that Watson found. Thus, Teach Watson plays an important role in improving the analysis model.

When users provide their feedback through Teach Watson, personal information is not stored anywhere. The following data is saved and used to enhance the processing model:
  • Tenant ID
  • Requirement text
  • User feedback about what to change in the requirement analysis
  • Industry (If user specifies this information.)
  • Business rule type
  • "Look for" information
Example:

Tenant ID: customerXTenantID
Requirement Text: "where the car is furnished with a GPS navigation system, the car shall enable the driver to mute the navigation instructions."
User feedback: "This is not passive in my context because of XYZ"
Industry: "Automotive"
Business Rule Type: "PASSIVE"
Look For: "is furnished"

This data is not acted on automatically by Watson. It is stored as input for the scoring model changes, and assessed by an IBM Requirement Subject Matter Expert. The data is deleted after it is reviewed.

The tenant ID is stored to be used for identifying and training a custom model based on the specific feedback from a particular enterprise.