"Novel" is the most overused word in research writing.
Earn it — or replace it with something specific.
Not "we also add X to the loss" but "we redefine the task to account for Y, which prior work ignores."
Combining A and B is novel only if the combination enables something neither A nor B alone can achieve — and you demonstrate this.
"We prove that X is equivalent to Y" or "We show that existing method Z fails because of condition W." Theory that explains behaviour is high-value novelty.
"We find that KG coverage, not model capacity, is the bottleneck for cold-start RS." A finding that changes how the field thinks is genuinely novel.
"We propose a novel framework that effectively integrates knowledge graphs with large language models for improved recommendation performance."
"Novel", "effective", "improved" — three content-free words. What specifically is new?
"We are the first to identify KG coverage sparsity as a failure mode of RAG-based RS, and propose a coverage-aware routing mechanism that reduces cold-start error by 23% without degrading warm-user performance."
First to identify + specific mechanism + specific result. Reviewers can verify all three.
Novelty test: Replace "novel" in your sentence with "first-of-its-kind" or "previously unknown." If it reads awkwardly — your claim is vague. If it reads naturally — your claim is specific enough.
Impact statements answer: "so what?" They appear in abstracts, introductions, and conclusions. Two levels:
What can researchers now do that they couldn't before? "Our analysis enables the design of retrieval policies that are provably better for sparse KGs."
What future directions does this make possible? "This opens a new research direction: KG-density-aware RS evaluation protocols."
Avoid: "Our work has important implications for future research." This says nothing. State the implication explicitly.
New formulation, non-obvious combination, theoretical insight, empirical finding. Know which one you're claiming.
Replace it with a specific claim. "First to identify", "We prove", "We show". Specificity is credibility.
State what your work makes possible — for practitioners, for researchers, for future work. Not just "we improve by X%."
Next: M6 · L6 — Responding to Reviews