Is Your Paper Ready for Expert Systems with Applications? A Readiness Verdict

These are the operating details that change a readiness decision, pulled from the publisher and the journal's published metrics.

Source: ESWA on ScienceDirect, ESWA guide for authors, Clarivate JCR 2024, accessed June 2026.

The journal's published timing signals are 5 days to first decision, 62 days to decision after review, 147 days to acceptance, and 7 days from acceptance to online. Community trackers report longer real-world averages, so treat the published medians as journal-level planning ranges, not a promise for your manuscript.

ESWA's desk screen is unusually fast, so presentation problems become rejection reasons rather than revision requests. These are the patterns a handling editor catches first:

Scope mismatch on application substance. The editor reads the abstract-and-Figure-1 pair to decide whether the application is real before the methodology takes over. If the application context is invisible, the paper is routed toward a methods venue within days.

An incomplete or stale comparison set. A benchmark table without current sister-venue baselines signals that the authors are not tracking the application-domain literature. Reviewers read this as a fairness problem, not just an omission.

Missing ablations. Reporting that a multi-component system performs well, without isolating which component does the work, leaves reviewers unable to credit the claimed contribution. Expect a major revision at minimum.

Statistical significance asserted without testing. A 0.3-point accuracy or F1 gain over the baseline, with no variance reported and no paired test across seeds or folds, is not a result a careful reviewer accepts at ESWA or its siblings.

Reproducibility gaps. Methods that describe evaluation splits in prose but provide nothing downloadable raise the question of whether the reported numbers hold, which extends the revision cycle.

Treat the abstract, Figure 1, benchmark table, ablation set, reproducibility files, and cover letter as one editorial intake package. Each component has a specific readiness test.

Abstract. Lead with the application domain and the decision it improves, then the method. The abstract should make a non-specialist editor understand why ESWA is the right venue within the first two sentences. A generic methods-first abstract is the first thing the desk screen flags.

Figures. Figure 1 must establish the application context: the data sample, the operational pipeline, the before-and-after decision setting. The architecture diagram belongs at Figure 2 or in supplementary material. This single change moves more papers from borderline to ready than any other.

Benchmarks and baselines. A fair comparison set is the load-bearing evidence at ESWA. Include at least two baselines from sister applied-AI venues published in the last 18 months, and state in the methods why each is the right comparator for your application claim. Stale general-AI baselines are read as a fairness failure.

Ablations. For every component you claim credit for, run an ablation that removes it and reports the delta. A system that wins overall but cannot show which part earns the win is not ready; reviewers will ask for the ablation, and adding it after submission costs a round.

Statistical significance. Report the spread across seeds or folds and a paired significance test for every headline metric. A single mean with no variance is the quiet readiness failure that survives author review but not reviewer scrutiny.

Reproducibility and code. Deposit code, evaluation splits, and trained model settings to a public repository (GitHub, Zenodo, or OSF) with a DOI, version tag, or commit hash before submission, and reference that identifier in the data availability statement. Replace "available on request" entirely.

Cover letter. The cover letter must argue why ESWA over its siblings, keyed to your contribution: knowledge-driven work points to KBS, soft-computing methods to Applied Soft Computing, engineering deployment to EAAI. A generic "ESWA is a leading journal" letter wastes the strongest fit argument you have.

Run an Expert Systems with Applications submission readiness check to confirm these components clear the bar before you upload.

In our pre-submission review work with applied-AI manuscripts targeting Expert Systems with Applications, four readiness patterns separate the drafts that survive the 5-day desk screen from the ones that come back fast. Each is testable against your own manuscript before you open the Editorial Manager upload slot, and each maps to a specific component the ESWA editor reads first.

Application invisibility in the abstract-and-Figure-1 pair. Across our ESWA pre-submission reviews, the most consistent readiness failure is an abstract that opens with a generic application claim ("we propose a deep learning approach for medical image classification") while Figure 1 is a model-architecture diagram rather than the application context. ESWA editors specifically screen that pair to decide whether the application is real before the methodology takes over.

When Figure 1 is an architecture schematic with no application-domain figure ahead of it, the manuscript reads as a methods paper that wandered into an application journal, and it is routed toward Pattern Recognition or Neurocomputing. The testable fix: make Figure 1 the decision setting, data sample, or operational pipeline, and demote the architecture to Figure 2.

Benchmark tables anchored on stale general-AI baselines. In our review work, ESWA-targeted manuscripts frequently build the comparison table from 2020 to 2022 baselines lifted from the method's original publication trajectory, then skip the recent application-domain methods published in Knowledge-Based Systems, Applied Soft Computing, or Engineering Applications of Artificial Intelligence in the last 18 months.

Because ESWA reviewers are application-domain specialists who track their own literature, a table missing current sibling-venue baselines reads as strategically incomplete and invites a reject-or-major-revision. The fix is to add at least two recent sister-venue baselines and justify each as the right comparator in the methods section.

Missing ablations on the claimed component. A pattern we see repeatedly across ESWA submissions is a multi-part system reported with strong aggregate numbers but no ablation isolating the contribution the authors are claiming credit for. The handling editor and reviewers cannot credit a contribution they cannot attribute, so the manuscript draws a major revision asking for exactly the ablation that should have shipped at submission. Before upload, ablate each claimed component and report the delta in the results table.

Reproducibility stated as "code available on request" with no significance testing. The fourth pattern is a data and code availability statement promising access "on request" rather than naming a public repository with a permanent identifier, often paired with headline metrics reported as a single mean with no variance or paired test across runs.

ESWA's editorial culture increasingly favors reproducible applied-AI work because performance claims are only checkable when the code, evaluation splits, and statistical spread are inspectable. Manuscripts that describe splits in prose but provide nothing downloadable, and report a thin accuracy gain with no significance test, draw extra reviewer scrutiny on whether the result holds.

The fix is to deposit code and splits to GitHub, Zenodo, or OSF with a DOI or version tag, and to report variance with a significance test for every headline number.

This page tells you what ESWA editors look for. A review tells you whether your specific draft passes the application-substance, baseline, ablation, and reproducibility screen before the upload commits you.

If the readiness matrix says your manuscript is not an ESWA paper, route on the reason, not on impact factor.

• Knowledge-Based Systems (Elsevier, JIF 8.8). The closest sibling when the contribution is knowledge-driven: reasoning, decision support, recommendation, or an intelligent system whose value is in how it represents domain knowledge. Reviewer reports can transfer via the Elsevier Article Transfer Service.

• Applied Soft Computing (Elsevier, JIF 6.6).

The right venue when the core method is fuzzy logic, evolutionary computation, swarm intelligence, or a hybrid soft-computing approach and the soft-computing technique itself is the contribution.

• Engineering Applications of Artificial Intelligence (Elsevier, JIF 8.0). For applications that live in an engineering or industrial system (control, manufacturing, energy, structural monitoring) where the deployment is the point.

• Neurocomputing (Elsevier).

For honest learning-method or neural-architecture contributions where the application was a thin demonstration; submit it as the methods paper it is.

• IEEE Access (IEEE, JIF 3.6). The speed-over-prestige route for correct, useful, reproducible work that is not novel enough to clear a selective applied-AI bar, with a roughly 4-to-6-week submission-to-publication timeline.

For a fuller map of the applied-AI landscape, see best machine learning journals.