Co-Analysis of Requirements Extracted from Multiple Sources

groanaberrantInternet και Εφαρμογές Web

2 Φεβ 2013 (πριν από 4 χρόνια και 4 μήνες)

185 εμφανίσεις

Analysis of Requirements Extracted from Multiple Sources

As companies grow, mature and re
organize there is a tendency of introducing
multiple and possibly conflicting systems to support software development. In
particular, this is true for larger companies with multiple organizational units. To
enable learning
and improvement of software development practices it is
necessary to be able to collect this information in order to be able to analyse and
understand the processes involved.

This thesis project aims to create a general framework for extraction of
ments and their meta
data from a multitude of diverse sources.
Requirements pose an additional challenge in that they are often in unstructured
(textual) form with fewer constraints since they need not be understood by
automated tools such as compilers or
test environments. Basically, requirements
are written, read and communicated only between humans which allows for a
larger variation in the way it is captured, documented and stored. We propose
that a general solution to this problem is to employ GUI auto
mation tools that
can act in a similar way to humans to interact with the systems where
requirements are kept and then extract information and send it off to an analysis
engine for further processing. The focus is on the extraction of requirements and
as m
uch of their meta data as possible. Two simple analysis tasks are added to
keep the project in focus and its expected outcomes clear. The project aims to
integrate with an existing probing/metric framework in development at the
Software Engineering departm
ent at Chalmers.

The thesis project is to be performed at Volvo Car Corporation which has a
complex requirements support structure with multiple sources for requirements
information. The project will be co
supervised by people from Volvo and people
from C



Analyze and describe the multiple requirements support systems in use at
Volvo and where and which requirements info can be found in them.


Study the Sikuli GUI automation tool and decide how to apply it to the
extraction of req info at

VCC. (see:



Create Sikuli scripts to extract req info from at least 3 different req
sources at VCC


Select and implement the targeted analysis tasks in Chalmers PROSE


Evaluate the strengths and weaknesses of the approach and its
generalizability to other organisations.


Students should preferably have good grades in SE and ReqEng courses.
Knowledge of Python or Ruby is a plus.


Robert Feldt, Chalm

and Joakim
Pernståhl, VCC,