1) Context-sensitive (Requirements-) Search:

Motivation: Plain keyword search is difficult to apply to large documents, since the number of false positives may be very large. For this purpose, a method shall be developed, that allows to formulate a query (in natural language) that takes the context in which any keyword is used into account.

Contact: Dr. Maximilian Junker

2) Visualization of Terminology:

Motivation: Choosing appropriate terminology is a difficult challenge in RE. Terminology often gets inconsistent. In addition, the topics described by terminology and their usage can hint at incompleteness of requirements (e.g. topics used in one part of the spec but not in another). In this thesis, we aim to:

1) Determine the use cases for terminology visualization for quality improvement or analysis.

2) Design a visualization to support one of the use cases of (1).

3) Evaluate the potential to improve the quality of requirements specifications.

Contact: Dr. Henning Femmer, Arjen Spaans

3) Detecting implicit causalities in natural language:

Motivation: Currently, Specmate can detect explicit causalities in natural language (e.g. “If it rains, the street will be wet”). The goal of this thesis is to evaluate techniques to identify implicit causalities. A possible solution could learn causal relationships between concepts from explicit formulations and use this to detect implicit causalities.

Contact: Dr. Maximilian Junker

4) Quality Metrics for IREB:

Motivation: Qualicen has a tool called Scout which analyses text of Requirements for potential quality defects similar to a spell checker. IREB is a standard which describes how good requirements management is done. The goal of this thesis is to extract rules for good requirements writing out of IREB that can be analysed with Scout.

Contact: Dr. Sebastian Eder