In particular, GUI prototypes can facilitate to mitigate and reduce misunderstandings between customers and developers, which may occur due to the ambiguity and vagueness of informal Natural Language (NL). However, employing high-fidelity GUI prototypes is more time-consuming and expensive compared to other simpler GUI prototyping methods. In this work, we propose a system that automatically processes Natural Language Requirements (NLR) and retrieves fitting GUI prototypes from a semi-automatically created large-scale GUI repository for mobile applications. We extract several text segments from the GUI hierarchy data to obtain textual representations for the GUIs. To achieve ad-hoc GUI retrieval from NLR, we adopt multiple Information Retrieval (IR) approaches and Automatic Query Expansion (AQE) techniques. We provide an extensive and systematic evaluation of the applied IR and AQE approaches for their effectiveness in terms of GUI retrieval relevance on a manually annotated dataset of NLR in the form of search queries and User Stories (US). We found that our GUI retrieval performs well in the conducted experiments and discuss the results.
