ANALYSIS OF SOFTWARE REQUIREMENTS

1
ANALYSIS OF SOFTWARE REQUIREMENTS

• S.Gnesi
• IEI-CNR Pisa
• Joint work with
• F.Fabbrini, M.Fusani, G.Lami

2
REQUIREMENTS ENGINEERING
3
Techniques for Producing SRS
4
SRS Quality Model

• Set of rules against which to evaluate a SRD
• Syntactic and semantic rules
• Document structure and sentence structure
  characteristics

5
SRS Quality Evaluation

• RSQ related Properties
• Non-Ambiguity the capability of a Requirement to
  have a unique interpretation.
• Completeness the capability of each Requirement
  to make reference to precisely identified
  entities.
• Understandability the capability of each
  Requirement to be fully understood when used for
  developing software.

6
SRS Quality Evaluation

• RDQ related Properties
• Completeness the capability of the Requirements
  Specification Document to avoid potential or
  actual discrepancies.
• Understandability the capability of the
  Requirements Specification Document to be fully
  understood when read by the user.

7
The Quality Model (I)
UNDERSTANDABILITY
8
The Quality Model (II)
TESTABILITY
CONSISTENCY
COMPLETENESS
9
EXAMPLES
10
QuARS Tool

• QuARS
• Quality Analyser of Requirements Specifications

11
(No Transcript)
12
(No Transcript)
13
NL SRS Evaluation Process
QuARS
14
Validation Results
S1 Business Application Functional
Requirements of a Transaction and Customer
Service (TACS) Check Cashing module S2
Space Software Application Functional
Requirements of a sub-system of a space vehicle
S3Telecommunication Application
Requirements Specification of a project aiming
for a new generation STM switches S4
Security Application Functional Security
Requirements for an Application Level Firewall
Protection Profile
Rate of defects occurrences on the total
requirements
Percentage distribution of defects types detected
15
Conclusions

• A Model for the syntactic quality of SRS
• uncomplete
• including indicators (metrics) numerically and
  automatically computable
• An automatic Tool providing support for the
  Quality Evaluation of SRS by means of calculation
  of the Quality Model metrics