A research gap is a defensible claim about what we do not yet know (or do not yet know how to do) for a specific question and context. In a thesis, a gap is only useful if you can support it with sources and show how your work will address it.
Different fields use different labels for gaps, and there is no single canonical taxonomy. Table Table 1 is a practical checklist you can use while reading: where do the current papers’ claims stop being justified?
| Gap pattern | What it means (in practice) | Example (logistics / OR) | How you address it in a thesis |
|---|---|---|---|
| Context / Setting | Evidence exists, but not under your constraints or decision environment | A method is tested in a toy setting, but your case has real-time constraints and noisy data | Explain why the context matters; adapt the approach; evaluate under your constraints |
| Model / Assumptions | Common assumptions likely drive the results, and they do not hold in your case | Papers assume deterministic demand, but your setting has stochastic arrivals and cancellations | Make assumptions explicit; relax/replace key ones; run sensitivity analysis |
| Method / Algorithm | The problem is known, but the available solution methods do not scale or do not apply | An exact formulation exists, but solvers time out on realistic instance sizes | Implement a solid baseline; justify design choices; report trade-offs (quality, time, robustness) |
| Evaluation / Benchmark | Claims exist, but comparisons are incomplete or weak | A new heuristic is reported without strong baselines, ablations, or a clear evaluation setup | Define baselines and metrics; compare fairly; document data, instances, and experimental protocol |
| Concept / Definition | Key terms are used inconsistently, so results are hard to compare | “Service level” is defined differently across studies | Define terms precisely; align measures with your research question; justify operational definitions |
| Integration / Workflow | Pieces exist separately, but no end-to-end workflow is evaluated in a realistic setting | Forecasting and routing are studied separately, but not as one pipeline | Build an end-to-end pipeline; show where errors propagate; discuss deployment constraints |
Turning Gaps into a Thesis Contribution
A “gap” is only useful if you can connect it to a concrete contribution and an evaluation plan. In most theses, the gap is not “nobody has ever studied X.” It is usually “existing work studies X under assumptions A/B/C, but my setting needs D/E/F.”
This mirrors a common move in academic introductions: summarize what is known, then show what is missing, then state what your work adds (Swales, 2004). Use Table Table 1 as a checklist while you read.
Swales’ CARS model (Create A Research Space) describes introductions as a sequence of rhetorical moves (Swales, 2004):
- Move 1: Establish a territory: show the topic matters; summarize what the field already knows.
- Move 2: Establish a niche: show what is missing, limited, or unresolved (this is where your gap statement lives).
- Move 3: Occupy the niche: state what you will do and what you will contribute (including how you will evaluate it).
A Practical Template
Use this structure to turn reading into a defensible thesis pitch:
- What exists: “Prior work on X focuses on …”
- What is missing: “However, Y is underexplored/unclear, especially in …”
- Why it matters: “This matters because …”
- What you do: “This thesis addresses Y by …”
- How you show it: “We evaluate by … (datasets, baselines, metrics).”
Common Pitfalls (and Better Alternatives)
- “There is no research on X.” → “Research on X is limited in settings with …”
- “No one considers Y.” → “Only a subset of studies considers Y, and results are mixed/limited because …”
- “This has never been done.” → “To our knowledge, there is limited evidence on … in …”
If you are doing a thesis with a strong literature component, keep your review traceable (search log, inclusion criteria). That makes your gap claims easier to defend (Page et al., 2021; Webster & Watson, 2002).