Information Technology Reference
In-Depth Information
stakeholders for symptoms of inconsistency, e.g., contradicting requirements; and (c)
Requirements
tracing
by identifying dependencies between requirements and artifacts
to support analyses for change impact and requirements coverage. Unfortunately,
ReqM suffers from the following challenges and limitations:
•
Incompleteness [7] of requirements categorization and conflict identification, in
particular, when performed manually.
•
High human effort for requirements categorization, conflict analysis and tracing,
especially with a large number of requirements [7].
•
Insufficient completeness [6] for conflict analysis and tracing with automated
approaches.
•
Tracing on syntactic rather than on concept level: requirements are often traced
on the syntactic level by explicitly linking requirements to each other. However,
requirements engineers actually want to trace concepts, i.e., link requirements
based on their meaning, which can be achieved only partially by information re-
trieval approaches like “keyword matching” [12] [13].
The use of semantic technologies seems promising to address these challenges:
Ontologies provide the means for describing the concepts of a domain and the rela-
tionships between these concepts in a way that allows automated reasoning [18].
Automated reasoning can support tasks for requirements categorization, requirements
conflict analysis, and requirements tracing.
In this paper, we propose OntRep, an automated ontology-based reporting
approach for requirements categorization, conflict analysis and tracing based on on-
tologies and semantic reasoning mechanisms. The main criteria for the evaluation are:
correctness and completeness of identified requirements conflicts, effort to develop a
project or domain ontology. OntRep aims at lowering the effort for requirements
management, while keeping high requirements consistency.
The OntRep approach automatically categorizes requirements into a given set of
categories using ontology classes modeled in Protégé and mapping the terms used in
the requirements to these classes. Further, OntRep analyzes the content of the re-
quirements and identifies conflicts between requirements. Therefore, conflict analysis
is not only based on traditional keyword-matching-approaches, but can also work
when different terminologies are used for requirements formulation.
We empirically evaluate OntRep with a real-life project at Siemens Austria, where
six project managers in two teams (a) categorized the requirements of the case study
project into a set of categories and (b) inspected the given project requirements to
identify conflicts between requirements. A requirements engineering expert provided
control data for all tasks. Then, we performed the same tasks with OntRep to compare
the effort necessary and the quality of results.
The remainder of the paper is organized as follows: Section 2 summarizes related
work on requirements categorization, conflict analysis, tracing, and natural language
processing technologies; Section 3 introduces the OntRep approach and motivates
research issues. Section 4 outlines the case study and Section 5 presents results.
Section discusses the results, concludes and suggests further work.
