Information Technology Reference
In-Depth Information
guided step-by-step through the requirements classification and conflict detection
process. The participants were sitting in one room without talking to each other. In the
team phase, the two teams (3 participants each) worked in separated rooms. The ex-
pert, as well as the OntRep tool user, also worked separately.
A5)
Background questionnaire:
before they started with the actual ReqM tasks, the
participants filled in the questionnaire.
A6)
Individual requirements categorization:
Then the participants read through the
23 given requirements. The participants individually categorized the requirements
into one or more of the given 8 categories. Each participant conducts the requirements
categorization individually. In addition, one Requirements Engineering expert also
does the categorization. The time needed by each participant is captured.
A7)
Individual requirements conflict analysis:
In addition to the 23 requirements
that had been categorized before, further elements were displayed (as rows below the
other requirements), namely: 11 constraints (technical and business), and 4 formal
documentation rules (documentation guidelines).
The participants again read through the task description and then had to identify
conflicts and enter them into the sheet. A conflict can have one of the following types:
conflict between requirements (CRR), conflict of a requirement with a constraint
(CRC), conflict of a requirement with a formal guideline, i.e., ill-formed requirement
(CRG). In total, the case study data contained: 5 conflicts of type CRR, 7 of type
CRC, and 10 of type CRG. After the evaluation of the manual approach, we again
have the 6 individual results. Again, one Requirements Engineering expert also con-
ducted the conflict analysis. The effort needed by each participant was captured.
A8)
Team requirements categorization & conflict analysis:
Afterwards, the par-
ticipants harmonized their individual results within 2 randomly assigned groups. Ef-
fort was captured for this task. The results are 2 team sheets.
A9)
Feedback forms:
filled in at the end by the participants.
A10, B6)
Evaluation of study results
: The manually created results of the expert
and the teams were then compared with the result generated by OntRep.
The process for the automated approach is:
B3)
Ontology preparation:
A tool expert created one ontology class in OntRep
(Protégé) for each category and then imported the given requirements from Trac as
CSV into OntRep.
B4)
OntRep requirements categorization:
The tool then executed the categoriza-
tion and generated a final result. We captured the effort to create the ontology classes
and to generate the final report.
B5)
OntRep requirements conflict analysis:
Then, we again provided the require-
ments as CSV-input to OntRep. Further, the tool expert had to model the constraints
as facts and the formal guidelines as rules in the ontology. We captured the effort for
this. Then, the tool executed the conflict identification and generated a final report.
Data Capturing Analysis and Statistical Evaluation.
Finally, we analyzed and
evaluated the following results: (a) 6 spreadsheets for requirements categorization and
6 spreadsheets for conflict analysis from each of the 6 individual participants, (b) 1
categorization spreadsheet and 1 conflict analysis spreadsheet from a requirement
engineering expert, and finally (c) 1 categorization spreadsheet and 1 conflict analysis
