Map of Scientific Uncertainty
The Scientific Question Compiler transformed a corpus on Empirical Research in Requirements Engineering (ORKG) into a structured representation of what is known, what is contested, and what is unknown. This is the inverse of a search engine: it maps the frontier, not the archive.
300
Documents
881
Claims
2
Contradictions
6
Gaps
7
Silence topics
13
Questions
Known
- Crowd-based Requirements Engineering (CrowdRE) centers on Elicitation, Analysis/Validation, Evolution, and Prioritization activities.
- Core utilities in CrowdRE are Crowd, Task, and Mechanism.
- User feedback classification in RE employs SVM, C4.5 Decision Tree, and Naive Bayes algorithms.
- Empirical RE studies in this domain are conducted at IEEE International Requirements Engineering Conferences.
- Threats to validity are routinely acknowledged, and data collection/analysis are standard practices.
- The research paradigm is reported as exploratory.
Controversial
- CrowdRE scope definition: limited to Elicitation/Analysis/Validation versus expanded to include Evolution/Prioritization.
- Research paradigm classification: exploratory versus explanatory.
- Integration of multimodal artifacts (video/ReqVidA) with traditional textual requirements minutes.
- Operational boundaries between crowd-sourced data generation and automated ML-based feedback classification.
Silence Map — implied but absent
- Replication and reproducibility protocols
- Comparative performance metrics of ML algorithms in RE contexts
- Participant demographics and crowd worker selection criteria
- Longitudinal impact of CrowdRE on software lifecycle
- Ethical considerations in crowd-based elicitation
- Specific datasets and preprocessing steps for user feedback classification
- Cost-benefit analysis of video vs. textual requirement documentation
Gaps (typed unknowns)
missing-mechanism
The corpus lists Crowd, Task, and Mechanism as utilities in Crowd-based RE but does not explain the operational interaction or workflow between these utilities during elicitation and validation activities.
severity 0.8
missing-link
The corpus discusses Crowd-based RE and ML-based user feedback classification in parallel but does not establish how crowd-sourced data feeds into or validates these classification models.
severity 0.7
unexplained-observation
The frequent reporting that research question answers are 'hidden in text' implies a methodological tendency in empirical RE to bury findings within narrative rather than stating them explicitly, yet this pattern is never critically examined.
severity 0.6
competing-theory
Conflicting claims about CrowdRE scope (limited to Elicitation/Analysis/Validation vs. expanded to include Evolution/Prioritization) imply an unresolved boundary definition for the paradigm that is never reconciled.
severity 0.75
missing-mechanism
While validity threats (internal, external, construct, reliability) are routinely acknowledged, the corpus omits the specific mitigation strategies or design controls employed to address them.
severity 0.85
missing-link
The introduction of video artifacts (ReqVidA) alongside textual minutes implies a multimodal requirements communication workflow, but the integration mechanism and comparative efficacy metrics are absent.
severity 0.65