The presentation suggests an improved way to produce high quality requirements at authoring stage, where the quality of the specification is calculated on the fly, while the analyst writes the requirements. The correctness of a single requirement, as well as the consistency and completeness of the specification is calculated and managed.
Tech Tuesday - Mastering Time Management Unlock the Power of OnePlan's Timesh...
From requirements management to requirements authoring - Innovate 2014
1. From Requirements Management to
Requirements Authoring –
An Experience Report from Airbus Group
Jean-Claude Roussel, Gerard Auvray -
Airbus Group
Juan Llorens – The REUSE Company
4. • 70% of defects are introduced during requirements phases
– It is important to write effective Requirements
• Experiences shown that about 25% of system Requirements are critical and can
grammatically be improved
– No Shall: 8 to 10%
– Forbidden words: 10 to 15%
– Subject, multiple objects, design: 15%
– Incorrect grammar: 50%, …
• Requirements error costs are high
– Fixing requirements after delivery may cost up to 100 times the one for fixing a requirement error
• Training, best practices and verifying requirements by reviews can help to get complete
and consistent requirements:
– But the process is costly and time consuming
• Reduce the number of iterations between System Engineers and sub-contractors and
improve the verification activities
Why doing Requirements Engineering ?
4
6. Survey on Industrial Practices:
Most common defects on Requirements
6
Study in the scope of RAMP project (Requirements Analysis and Modeling
Process) in partnership with Airbus Group, RENAULT, EDF, ADN, CORTIM,
ENSTA, IRIT, PARIS 1 UNIVERSITY
(end 2010 over 22 industrials in several domains worldwide: interviews and questionnaires)
06/20/14
Requirements are
mainly written in natural
language
QUALITY
ISSUES !
6
7. A perfect…bad requirement !
Whatever the context, the system should be able to
switch-off as quickly as possible, to restart without
inducing any perturbation and at the same time to
inform the operator in a friendly and understandable
manner…
Need to perform a Requirement Quality Analysis against all the set of
the system Requirements to improve the requirement quality before any
delivery.
Requirement not SMART (Specific, Measurable, Attainable, Realisable, Traceable)
Ambiguous (What System ?)
Several muddled requirements (switch-off, restart, inform)
No use of shall
Not measurable (quickly, same time,,..)
Not testable (friendly, understandable,..)
…
7
8. From Requirements Management
to Requirements Authoring
Requirements Quality Control:
Huge review effort for quality verification
My tailor is rich
I should need RAT
Requirements Writing:
Working with requirements inside DOORS
8
9. Requirements Quality Control:
Huge review effort for quality verification
Requirements Quality Assurance:
Authoring “right the first time” on top of DOORS
My tailor is rich
I should need RAT
Requirements Writing:
Working with requirements inside DOORS
Requirements Quality Control:
Smaller review effort for quality verification
From Requirements Management
to Requirements Authoring
9
10. A first answer :
Requirements Quality Analysis (RQA) Tool
– Tool delivered by TRC (The Reuse Company)
– Completely linked with DOORS Tool
– Focusing on Syntactic analysis (DQA) and Semantic analysis (RQA V4.1)
– Tool evaluation
– Airbus Group Innovations
– Airbus Defense and Space
– Close cooperation with TRC in order to improve the tool
– Tool deployment
– Limited in Airbus Defense and Space
– R&T project focusing on how to improve the requirements quality
analysis process
– Operationally deployed in Airbus (Commercial Aircraft)
• A320neo Cabin Flex
10
11. From Requirements Analysis to Authoring
So far RQAV3 was a verification tool:
– Checking the quality of individual requirements once they are written
– Based on +25 metrics
– Checking the consistency of the overall specification
RQA has evolved towards the Requirements Quality Suite (RQS) inV4
– Supports the CCC approach (Correctness, Consistency and Completeness)
– V4 is mainly focused on “authoring”
– Includes RQA, Requirements AuthoringTool (RAT) and KM
RAT:“The SMARTer* way to create requirements”:
An assistant to Engineers during the authoring process
Including a wide set of customizable boilerplates
Helping Engineers during authoring with an intellisense system
* SMART stands for Specific, Measurable, Attainable, Realisable, Traceable
11
13. The Requirements Quality Suite (RQS)
The Requirements Quality Suite (RQS) intends to tackle requirements quality
management by offering a set of tools and processes.
RQS models requirements quality using the CCC approach (Correctness,
Consistency and Completeness)
Requirements Quality Analyzer (RQA):
to setup, check and manage the quality of a
requirements specification.
Requirements Authoring Tool (RAT):
to assist authors in the process of creating
requirements with the right level of quality
knowledgeMANAGER (kM):
to manage knowledge around a
requirements specification:
-the ontology it is based on
-the structure of the requirements to be used in the
project
-the communication between authors and domain
architects.
13
15. Correctness :
Individual requirement supported metrics
– Size
– Readability
– Conditional vs. imperative
sentences
– Active vs. passive voice
– Optional sentences
– Ambiguous sentences
– Subjective sentences
– Implicit sentences
– Abuse of connectors
– Negations
– Speculative sentences
– Use of false friends
– Design terms
– Flow terms
– Number of domain nouns and
verbs
– Acronyms
– Hierarchical levels
– Volatility
– Number of dependences
– Forbidden Words
– Standard Requirement (match
pattern)
Page 15
15
16. Consistency : Supported Metrics
Consistency applies to requirements sets
–Redundant requirements: Several requirements expressing the same
need at the same level of abstraction.
–Inconsistent units: Different requirements in the same
module/block/project uses different metric units.
–Inconsistent Content: Different requirements present value restrictions
that are not compatible.
16
17. Completeness: Supported Metrics
Completeness applies to requirements sets
–Missing requirements: Lacks the existence of requirements expressing
the same need at the different level of abstraction in different modules/blocks
of the same project.
–Missing Links Lacks the existence of links between requirements
expressing the same need at the different level of abstraction in different
modules/blocks of the same project.
17
18. RAT : Requirements Authoring Tool
• Main features:
– Assisting authors while they’re writing requirements
– Following a agreed upon set of patterns
• Other (on the fly) features:
– Quality assessment (correctness based
on individual metrics) on the fly
– Consistency analysis on the fly
– Missing links on the fly
– Inconsistent units analysis on the fly
18
19. • Author assistance on the fly (typing requirement)
Pattern choicePattern choice
Valid paths to fulfill the selected patterns
Ontology terms to keep
fulfilling the selected patterns
RAT : Requirements Authoring Tool
19
20. • Author assistance on the fly (done requirement)
Pattern fulfilledPattern fulfilled
RAT : Requirements Authoring Tool
20
22. • Semantically similar requirements on the fly
There’s a requirement in the SKB very similar to the writing requirementThere’s a requirement in the SKB very similar to the writing requirement
RAT : Requirements Authoring Tool
22
23. • Inconsistent measurement units on the fly
There’s a requirement in the SKB conflicting with the writing requirementThere’s a requirement in the SKB conflicting with the writing requirement
RAT : Requirements Authoring Tool
23
24. Knowledge Manager
• Main features:
– Vocabulary management
– Thesaurus management
– Requirements Patterns management
– Semantic indexing and retrieval services
• Ontology construction:
– By reusing existing vocabulary, thesauri, ontologies
– By creating a semantic graph out of existing
documentation
– By “elicitating” with the help of experts
– By attending the suggestions provided by end users
24
26. Controlled vocabulary: valid
terms, forbidden terms…
Optionally can include a
Glossary (description for every
term)
Taxonomy: terms hierarchically
organized
Thesaurus: hierarchies,
associations, synonyms…
Light Ontology: syntactic and
Semantic groupings for Terms
and Actions (verbs). Domain
terms and verbs
Full-Ontology: patterns and requirements
structures
Pattern Management
Full-Ontology: boilerplates and
requirements structures
Knowledge (Ontology)
Management
Page
26
Knowledge Manager
27. Thank You!
Your Feedback is Important!
Access the Innovate agenda tool to complete your
session surveys from your smartphone, laptop or
conference kiosk.
27