Abstract: Requirements cannot be collected, but must be elicited from people’s tacit and systems ’ embedded knowledge. It is a proven approach to use existing systems and manuals as a source. Literature suggests using background reading to elicit requirements from domain descriptions and manuals. Besides content analysis, no concrete technique describing how to actually conduct background reading is available in literature. In this paper, we evaluate the usage of template analysis as a technique for background reading. We applied template analysis in a project to extract requirements from 35 success stories about process performance management. We found template analysis to be very useful not just for eliciting requirements and creating a shared understanding of the studied domain, but also for helping new employees to get familiar with the domain. We also formulated competence questions to document and communicate requirements, but this did not prove helpful and hence we would not recommend it. 1

Requirements elicitation and analysis remains a stubbornly intractable problem to automate. This paper looks at the use of natural language analysis techniques in the requirements capture process. It begins with a review of past work and then develops an algorithmic approach to analysing and restructuring natural language text. We begin by marking the text up as parts of speech, before we restructure this mark-up into subject-verb-object clauses. After a user-assisted pronoun replacement step, a relatively comprehensive UML class diagram can be generated. Further analysis then allows us to generate use cases. To illustrate the algorithm in practice we use a case study.