Speculative requirements: Automatic detection of uncertainty in natural language requirements

Abstract

Abstract-Stakeholders frequently use speculative language when they need to convey their requirements with some degree of uncertainty. Due to the intrinsic vagueness of speculative language, speculative requirements risk being misunderstood, and related uncertainty overlooked, and may benefit from careful treatment in the requirements engineering process. In this paper, we present a linguistically-oriented approach to automatic detection of uncertainty in natural language (NL) requirements. Our approach comprises two stages. First we identify speculative sentences by applying a machine learning algorithm called Conditional Random Fields (CRFs) to identify uncertainty cues. The algorithm exploits a rich set of lexical and syntactic features extracted from requirements sentences. Second, we try to determine the scope of uncertainty. We use a rule-based approach that draws on a set of hand-crafted linguistic heuristics to determine the uncertainty scope with the help of dependency structures present in the sentence parse tree. We report on a series of experiments we conducted to evaluate the performance and usefulness of our system.

Improving agile requirements: the Quality User Story framework and tool

“…They apply their tool to 11 full-text requirements documents and find that it performs reasonably well in identifying uncertainty cues with F-scores of 62 % for auxiliaries, verbs, nouns, and conjunctions. On the other hand, it under-performs in identifying the scope of detected uncertainty causing the overall F-score to drop to 52 % [56].…”

A Unified Framework to Identify and Extract Uncertainty Cues, Holders, and Scopes in One Fell-Swoop

“…Some researchers cast uncertainty detection as a token sequence labeling problem, and then use hand-cra ed rules to extract scopes Ørelid et al (2010);Yang et al (2012); Apostolova et al (2011);Velldal et al (2010). Others de ne both uncertainty detection and scope extraction as token sequence labeling problems; yet use separate feature sets for each task or do not use the output of one task to inform the other Zhao et al (2010).…”

“…None of the research that tackles the two uncertainty-related tasks simultaneously uses a uni ed framework. For example, Ørelid et al (2010), Yang et al (2012), Apostolova et al (2011), and Velldal et al…”

Uncertainty handling in goal-driven self-optimization – Limiting the negative effect on adaptation

“…This study is more qualitative (e.g., are there more alternatives to achieve a goal) where ours is more quantitative (e.g., is the contribution link underestimated). Yang et al (2012) proposed an approach to detect the uncertainty in natural language requirements. This study handles requirements uncertainty at the language level where ours is at the model level.…”

“…While design-time uncertainty handling in requirements has been widely addressed through better approaches in elicitation, disambiguation and inconsistency check (Liaskos et al, 2012;Yang et al, 2012;Arora et al, 2012), these approaches more or less involve human intervention. Hence, runtime adaptations through unsupervised feedback loops present little time to apply them, while quickly responding to changes in the environments.…”