Requirements Engineering - Quality assurance
•
2 j'aime•1,458 vues
16th slide set of CECS 542 Quality assurance for requirements documentation Complete course: http://foss2serve.org/index.php/Requirements_Engineering,_CSU_Long_Beach,_Penzenstadler
Recommandé
Recommandé
Contenu connexe
Tendances
Tendances (20)
Similaire à Requirements Engineering - Quality assurance
Similaire à Requirements Engineering - Quality assurance (20)
Plus de Birgit Penzenstadler
Plus de Birgit Penzenstadler (15)
Dernier
Dernier (20)
Requirements Engineering - Quality assurance
- 3. Learning Goals • Founda@ons of quality assurance – Quality criteria for RE – Construc@ve and analy@cal Quality Assurance • QA for Artefacts • Techniques for Quality Assurance Dr. Birgit Penzenstadler 3 Context Layer System Layer Requirements Layer Stakeholder Model Objectives & Goals Constraints & Rules ! ! ! ! ! Data Model E A A A E System Vision Functional Hierarchy Architecture Overview System Function Model Fun 1 Fun 2 Component Model C C Data Model E A A A E Behaviour Model Business Case Deployment Requirements System Constraints Domain Model Service ModelUsage Model Quality Requirements Risk List Project Scope Process Requirements Glossary Glossary Glossary
- 4. Overview: Quality Assurance • Mo@va@on and Terminology • Quality of Requirements Documents • Principles of Quality Assurance • Techniques for construc@ve QA • Techniques for analy@cal QA Dr. Birgit Penzenstadler 4 K Rayker, stock.xchng
- 5. 16 11 9 1 7 7 5 3 3 1 5 1 4 3 0 2 1 1 0 31 22 22 20 16 13 13 12 11 11 9 9 8 8 5 3 3 2 2 0 5 10 15 20 25 30 35 Incomplete / hidden reqs. Moving targets Time boxing Separa@on reqs. from known solu@ons Underspecified reqs. Communica@on flaws to customer Inconsistent reqs. Communica@on flaws in team Missing traceability Gold pla@ng Unclear non-func@onal reqs. Terminological problems Insufficient support by customer Unclear responsibili@es Vola@le domain Weak access to customer needs Insufficient support by project lead Technically unfeasible reqs. High degree of innova@on vs. need for formal acceptance Overall frequency Cause for project fail Recap: Challenges in prac@ce 5 Requirements are o.en... • Incomplete • Not agreed upon • Contradictory • Not measurable • Unfounded • Irreproducible Mendez et al. Naming the Pain in Requirements Engineering – The NaPiRE-Report
- 7. Terminology in the context of quality assurance in RE Quality defect • Incorrect (invalid) requirement: Requirement that does not reflect the inten@on of the stakeholder (in the sense of „validity“) • Quality defect: Requirement that can be valid, but has qualita@ve defects, e.g. missing measurability, low understandability, contradictory, ... • Interrela@on of those two: – Incorrect requirements are ojen hidden due to quality defects – Correctness of requirements ojen viewed as quality criterium Valida7on and Verifica7on • Valida@on: Check of requirement w.r.t. correctness (it‘s a valid requirement, meaning it represents the inten@on of the stakeholder) • Verifica@on: Check of system w.r.t. fulfillment of requirements • Both are part of QA Dr. Birgit Penzenstadler 7
- 8. Quality Assurance in Requirements Engineering • Def. QA in RE: Applica@on of systema@c measures for iden@fying quality defects and assuring the quality of the requirements specifica@ons. à Check of quality criteria, e.g.: § Correctness § Completeness § Consistency § Traceability § ... (see following slides) à The examina@on can be conducted construc@ve or analy@cal using a formal procedure. 8
- 9. Overview: Quality Assurance • Mo@va@on and Terminology • Quality of Requirements Documents • Principles of Quality Assurance • Techniques for construc@ve QA • Techniques for analy@cal QA Dr. Birgit Penzenstadler 9 K Rayker, stock.xchng
- 10. Recap: Requirements, documents, artefacts, repositories • A requirement is a demanded characteris@c of a system or process. • We dis@nguish – Syntac@c representa@on: Text, table, diagram, formula – Seman@c representa@on: Content – informal or formal • A requirements document or artefact – Contains a number of requirements – Has a structure • Delimita@on: A requirements repository (database for requirements) serves for storing large sets of requirements (and requirements artefacts) 10
- 11. QA of requirements documents • Quality of requirements documents is crucial for project success. à Why is that? What is based on them? • We need specific procedures for QA • Relevance of the quality criteria needs to be determined by the further use of the documents. UCI winter 2014 Dr. Birgit Penzenstadler 11
- 12. Focus of quality assurance in RE Perspec@ves in QA in RE • We dis@nguish the quality of – Requirements documents / artefacts – Sets of requirements / statements – Individual requirements – Systems 12 Context Layer System Layer Requirements Layer Stakeholder Model Objectives & Goals Constraints & Rules ! ! ! ! ! Data Model E A A A E System Vision Functional Hierarchy Architecture Overview System Function Model Fun 1 Fun 2 Component Model C C Data Model E A A A E Behaviour Model Business Case Deployment Requirements System Constraints Domain Model Service ModelUsage Model Quality Requirements Risk List Project Scope Process Requirements Glossary Glossary Glossary
- 13. Overview: Quality Assurance • Mo@va@on and Terminology • Quality of Requirements Documents • Principles of Quality Assurance • Techniques for construc@ve QA • Techniques for analy@cal QA Dr. Birgit Penzenstadler 13 K Rayker, stock.xchng
- 14. Analytical QS Depending on the quality criteria, responsible for checking are: • Project team members with domain knowledge during elabora@on of the requirements, e.g. „correctness“ à this is called construc@ve QA • External/neutral quality responsibles who perform checks, e.g. „traceability“ and „understandability“ à this is called analy@cal QA à Which measures can you think of for performing either of these? Constructive QS Principle of construc@ve and analy@cal QA 14 Context Layer System Layer Requirements Layer Stakeholder Model Objectives & Goals Constraints & Rules ! ! ! ! ! Data Model E A A A E System Vision Functional Hierarchy Architecture Overview System Function Model Fun 1 Fun 2 Component Model C C Data Model E A A A E Behaviour Model Business Case Deployment Requirements System Constraints Domain Model Service ModelUsage Model Quality Requirements Risk List Project Scope Process Requirements Glossary Glossary Glossary Project team Quality responsible
- 16. Overview: Quality Assurance • Mo@va@on and Terminology • Quality of Requirements Documents • Principles of Quality Assurance • Techniques for construc@ve QA – Reference Models – Quality criteria according to IEEE 830 – Linguis@cs – Guidelines • Techniques for analy@cal QA Dr. Birgit Penzenstadler 16 K Rayker, stock.xchng
- 17. Reference Models • AMDiRE • IEEE 830 • Cockburn template • UML2 standard Dr. Birgit Penzenstadler 17
- 18. Construc@ve QA in RE: Examina@on of Quality Criteria acc. to IEEE 830-1998 • Completeness: Systema@cally run through all cases (You can only find incompleteness, not guarantee completeness!) • Consistency: Relate all to one system model • Unambiguity: Check phrasing • Correctness: Valida@on • Structuredness: Examine structure • Traceability: Are requirements sufficiently linked? • Changeability: Can expected changes in requirements be conducted efficiently? • Understandability: Check formula@ons • Agreed upon: Check with stakeholders 18
- 19. Checklist: Ques@ons and Criteria for ... • Completeness: • Consistency: • Unambiguity: • Correctness: • Structuredness: • Traceability: • Changeability: • Understandability: • Agreed upon: 19
- 20. Checklist: Ques@ons and Criteria for a spreadsheet with requirements • Completeness: Read through, are there s@ll open ques@ons? Lead engineer, stakeholder, customer • Consistency: Check for requirements conflicts • Unambiguity: language check, let someone else read it whether it could be misunderstood • Correctness: check with stakeholder • Structuredness: adheres to template / outline, has a breakdown that makes sense • Traceability: links between requirements in spreadsheet, across versions of spreadsheet • Changeability: check dependencies between requirements (high degree of dependencies means low changeability) • Understandability: let someone else read it whether it‘s easy to understand • Agreed upon: check with stakeholder 20
- 21. Linguis@cs in RE • Classifica@on of linguis@c quality defects – lexical/ontological (what does „green“ mean?) – syntac@c (“I saw the man on the hill with a telescope”) – seman@c (“All persons have a unique na@onal insurance number“) – pragma@c (“The trucks shall treat the roads before they freeze“) – weak phrases: (“as soon as possible“) – Omission or generaliza@on • Syntax paperns – [when?] [under what condi@ons?] THE SYSTEM SHALL | SHOULD | WILL <process> <thing to be processed> [<process detail>*] 21
- 23. Phrasing: Do‘s and Don‘ts 1. The system shall respond as fast as possible. In 90% of all cases, the system shall respond to all queries within 3s. 2. Students take 10 courses per semester. Students take 1000 courses per semester. Every student takes 10 courses per semester. 3. Shortly before the due date the medium is extended, unless somebody else reserved it. Three days before the due date, the system checks whether the medium has been reserved. If not reserved, the system extends the lending period. 23
- 24. Guidelines & Checklists How would you write a guideline and a checklist for … ? Team up! • Stakeholder Model • Goal Model • System Vision • Usage Model • Non-func@onal Requirements Dr. Birgit Penzenstadler 24
- 25. Overview: Quality Assurance • Mo@va@on and Terminology • Quality of Requirements Documents • Principles of Quality Assurance • Techniques for construc@ve QA • Techniques for analy@cal QA – Checklists – Quality Gates – Fagan Inspec@ons Dr. Birgit Penzenstadler 25 K Rayker, stock.xchng
- 26. Exemplary Check list according to Lamsweerde (1/2) • General check: Must be clear "what, who, when, where“ • Defect-based criteria (as in IEEE 830): – ContradicJon – Inadequacy – Unmeasurability – Unfeasibility – Poor structuring – ... • Quality-specific criteria: – Is there any unspecified response in this operaJon to not receiving an expected input value, or receiving it too early or too late? – Does the logical OR of the input condiJon on this operaJon form a tautology? – ... 26
- 27. Exemplary Check list according to Lamsweerde (2/2) • Domain—specific criteria: typical issues in the parJcular domain • Content-related criteria: – Templates • all fields filled • idenJfier user consistently • statement type correct • ... – Graphical nota@ons • data flow consistent • ER Diagram declaraJon • ... – Formal specifica@ons 27
- 28. Quality Gates Specific milestone in a sojware project that checks • Content: The „usual“ quality criteria: Completeness, consistency, ... • Documenta@on: Compliance with format, understandability, unambiguity, ... • Accordance: Every requirement agreed upon, conflicts resolved, ... 28 Role Model Process Model Project Scope defined System Specification accepted Business Analyst Requirements Engineer System Architect Architecture Overview defined Requirements Specification accepted System Vision defined Context Specification accepted Context Specification System Specification Requirements Specification
- 29. Fagan Inspec@on The term Fagan inspecJon refers to a structured process of trying to find defects in development documents. It includes the following phases: • Planning: Moderator plans review process • Overview: Author describes the background of the document under inspec@on • Prepara@on: Every reviewer examines the document in order to find defects. • Inspec@on mee@ng: A specific reader walks through the document chapter by chapter, and the inspectors point out found defects. • Adapta@ons: The author of the document corrects the found defects according to the ac@on plan agreed upon in the mee@ng. • Follow-up control: The inspectors check whether the defects were fixed correctly. 29 Planning Overview Preparation Inspection Meeting Adaptation Follow-up Control
- 30. IEEE 730 – 2014: IEEE Standard for Sojware Quality Assurance Processes Dr. Birgit Penzenstadler 30
- 31. IEEE 730 – 2014: IEEE Standard for Sojware Quality Assurance Processes Dr. Birgit Penzenstadler 31
- 32. IEEE 730 – 2014: IEEE Standard for Sojware Quality Assurance Processes Dr. Birgit Penzenstadler 32
- 33. IEEE 730 – 2014: IEEE Standard for Sojware Quality Assurance Processes Dr. Birgit Penzenstadler 33
- 34. IEEE 730 – 2014: IEEE Standard for Sojware Quality Assurance Processes Dr. Birgit Penzenstadler 34
- 35. Take-away: QA • Defini@ons – Quality Assurance – Quality Defect • Construc@ve QA – Guidelines and criteria – Reference models • Analy@cal QA – Quality gates – Fagan inspec@on – Checklists • IEEE 730 Std for SQA Dr. Birgit Penzenstadler 35 Context Layer System Layer Requirements Layer Stakeholder Model Objectives & Goals Constraints & Rules ! ! ! ! ! Data Model E A A A E System Vision Functional Hierarchy Architecture Overview System Function Model Fun 1 Fun 2 Component Model C C Data Model E A A A E Behaviour Model Business Case Deployment Requirements System Constraints Domain Model Service ModelUsage Model Quality Requirements Risk List Project Scope Process Requirements Glossary Glossary Glossary