Journals Assistant · User Preference Reference
Journals Assistant · User Preference Reference
AI-driven journal intelligence
A practitioner’s reference for precision journal discovery. It documents the nine Preference Impact Indicators: what each one constrains, the cue words that activate it, the values it accepts, and the phrasings that silently fail.
A preference is extracted only when a cue word signals it. A country, publisher, subject or language that merely appears inside a journal title, or inside your own research topic, never becomes a filter. This is the cue-signal principle, developed in §15 and restated wherever an indicator depends on it.
Type a query as you would in the Journals Assistant. The workbench separates instruction from subject matter using the rules in this guide.
Front matter
This guide documents the preference layer of the i-Future AI Journals Assistant: the vocabulary you add to a research question to constrain, rank and filter the journals the system returns.
The Journals Assistant accepts ordinary language. You need no operators, brackets or Boolean syntax. You do need to know which words the extraction layer treats as instructions and which it treats as subject matter, because the same word can be either. Frontiers is a publisher when it follows published by, and part of a journal title in Frontiers in Bioengineering. The sections that follow teach that distinction systematically.
A trigger phrase. Its presence tells the system that a filter is intended.
A phrasing the extraction layer resolves unambiguously.
A phrasing that is silently misread, dropped or over-applied.
Blue, violet and teal mark the three indicator families introduced in §5.
Beside an example, shows how the workbench decomposes it: the filters it yields and any warnings.
§ numbers identify sections of this guide. Indicator numbers identify the nine preferences and follow the order of the source specification. Indicators 1 to 9 are documented in §6 to §14, so Indicator 2 (quartile ranking) is §7 and Indicator 9 (article processing charge) is §14.
This guide is written for researchers, postgraduate candidates, research-office staff and editorial advisers who use the Journals Assistant to identify submission targets. It assumes familiarity with scholarly publishing and defines each term in the Glossary (§22). It does not cover the other four i-Future AI models.
The nine preference settings documented here are the Preference Impact Indicators, called indicators throughout and numbered one to nine. Earlier editions called them dimensions; the two terms are interchangeable. The numbering is stable across editions, so cross-references remain valid.
Front matter
The guide is written to be read three ways. Choose the route that matches how much time you have and what you need from the system today.
Read the query anatomy, then copy a pattern from the worked examples and substitute your own topic.
Work through the nine indicators in order. Each is self-contained.
Use the master cue-word index, or search, to jump from a phrase you want to use to the rule that governs it.
Part I
Where the Journals Assistant sits in the workspace, the three forms your research input can take, and the structural grammar every preference query follows.
§1
Five assistants share one interface. Preferences behave differently in each, so the first act of any session is choosing the right one.
i-Future AI presents a single question box backed by five specialised models. The model selector sits at the top left of the workspace, beneath the product name. The assistant you choose determines what the system searches and which preference vocabulary it honours. Everything in this guide applies to the Journals Assistant.
| Assistant | What it is for | Preference vocabulary |
|---|---|---|
| Research Topics Assistant | Discovering and refining candidate research topics | Topic-level |
| Research Topics Expert | In-depth expert analysis of a chosen topic | Topic-level |
| Gen Research Assistant | General-purpose generation: explaining a concept, drafting a section, outlining a method | None |
| Journals Assistant | Finding and evaluating journals for a manuscript | The nine indicators in this guide |
| Smart Search | Synthesising across a curated research collection | Collection-level |
Table 1.1The five i-Future AI models. Preference clauses written for the Journals Assistant have no effect in the other four.
Once the Journals Assistant is active, four cards appear below the question box. They are shortcuts, not modes: each pre-fills a pattern of question, and you may ignore them and type your own query.
| Card | What you supply | What you receive |
|---|---|---|
| Find the Right Journal for Your Research | A topic or short phrase, plus preference clauses | Ranked candidate journals |
| Enter Manuscript Text | Abstract or full manuscript text, plus preference clauses | Journals matched on content, then filtered |
| Journal Risk and Quality Assessment | A named journal or a shortlist | Quality and risk indicators for each |
| Evidence-Based Manuscript Validation | Manuscript text | Claim- and evidence-level checks |
Table 1.2Journals Assistant task cards. Preference clauses matter for the first two; the other two evaluate material you have already identified.
The cards work best in order. A defensible submission decision moves through three stages, and preferences shape only the first.
Cards 1 and 2
Candidate journals matched to your topic or manuscript, narrowed by the indicators in Part II.
Card 3
The shortlist interrogated for indexing withdrawal, editorial opacity and other risk signals.
Card 4
The manuscript examined for whether it supports the claims it makes.
Figure 1.2From discovery to readiness. Preferences shape the first stage; they cannot substitute for the second.
§2
Topic, abstract or full manuscript. The choice governs how much semantic signal the matcher has to work with, and therefore how much filtering you should apply on top.
Every Journals Assistant query begins with a statement of what your research is. You may express it at any of three levels of resolution, with or without preference clauses. The levels are not ranked; each suits a different moment in the publication cycle.
| Input mode | When to use it | Behaviour and trade-off | Working ceiling |
|---|---|---|---|
| Research topic a phrase or short line | Early scoping; testing whether a field has viable outlets before committing to a study | Broadest recall. The matcher has few terms to work with, so the returned set is wide and heterogeneous. Apply preference clauses generously: filtering is what gives a topic query its precision. | 5–6 indicators |
| Research abstract 150–300 words | Manuscript drafted or near-complete; the study’s contribution is settled | The most efficient mode for most users. An abstract carries enough methodological and topical signal for meaningful semantic matching and is quick to paste. Two or three clauses usually suffice. | 3–4 indicators |
| Full manuscript complete text | Final target selection; resubmission after desk rejection | Highest-fidelity matching, including method, instrumentation and reference profile. Apply preference clauses sparingly: the content signal is already strong, and heavy filtering will empty the result set. | 2–3 indicators |
Table 2.1Input modes and the filtering posture each implies. The working ceilings are the rule of thumb restated in §16.
The richer your research input, the fewer preference clauses you need, and the more likely it is that adding them will over-constrain the search. If a full manuscript returns nothing, remove filters before you rewrite the input (§18).
Abstracts and manuscripts may be pasted into the question box or attached with the paperclip control at its right edge. Attached text is treated exactly like pasted text. Type preference clauses into the question box alongside the attachment rather than inside the document: a filter written inside an uploaded manuscript is read as manuscript content, not as an instruction. This is the cue-signal principle at work (§15).
§3
One research input, followed by any number of filter clauses, in any order, in ordinary prose.
A Journals Assistant query has two structural components. The first is your research input: the topic, abstract or manuscript described in §2. The second is a sequence of filter clauses, each introduced by a cue word that names the indicator it constrains. There is no fixed order, no punctuation requirement and no limit on the number of clauses beyond the practical one discussed in §16.
The extractor scans for cue words. Each cue word and the value that follows it are lifted out as a filter; everything that remains is the research input.
Demonyms become country names, colloquial language names become standard ones, currency figures become plain numbers, and values are cast to Title Case (§20).
The remaining text is matched against the journal corpus on topic, scope and aims, not on keyword overlap alone.
Filters combine with AND across indicators and OR within one. A journal must satisfy every indicator you name, but any one of the values you list inside it (§16).
Moving a clause earlier or later in the sentence changes nothing. Omitting its cue word changes everything: the value collapses back into your research input and silently widens the search instead of narrowing it.
Palm oil Indonesia searches for journals about palm oil in Indonesia. Palm oil from Indonesia searches for palm-oil journals published in Indonesia.
§4
The same research interest at three levels of constraint. Read these before Part II; they make the individual indicators easier to place.
Appropriate for first orientation in an unfamiliar field. Expect a long, heterogeneous list spanning tourism studies, Islamic studies, hospitality management and food science.
The working default for most submission decisions. Q1 and Q2 are alternatives within one indicator: a journal qualifies by meeting either.
Eight indicators at once, above the working ceiling for a topic (§16). Queries of this density often return nothing, and that result is informative: it tells you the constraint set is empty. Relax in the order given in §18.
An empty return from a well-formed Level 3 query is a finding, not an error. It states that no journal in a corpus of more than 206,000 titles satisfies all eight conditions at once. The productive response is to identify which condition does the excluding. It is usually the conjunction of a national constraint with a quartile constraint, since high-quartile titles concentrate in a small number of publishing countries. Level 3 also pairs country with language, which Table 16.1 marks as redundant. Decide which of the conflicting conditions matters more to you, and relax the other.
Part II
Each indicator is documented in the same format: what it constrains, the cue words that activate it, the values it accepts, worked examples, and the failure modes that silently produce the wrong result.
§5
The nine indicators answer three different questions about a journal. Knowing which question you are asking is the fastest way to choose the right filter.
The source specification numbers the indicators one to nine in the order they were implemented. This guide keeps that numbering so references stay stable, but the order hides a structure worth making explicit: every indicator belongs to one of three families, and the families differ in how sharply they cut.
Where the journal is recognised and how it ranks. These are the strongest filters: they map directly onto the criteria most institutions use to assess an output.
How the journal operates. These govern your experience as an author: how long publication takes, how review is conducted, and what publication will cost.
Who publishes it and from where. Useful for regional visibility, institutional mandates and language reach, and the most likely to over-constrain.
Specify one filter from Family A, one from Family B, and only then consider Family C. Family A establishes that the journal will count for your assessment purposes. Family B establishes that its process suits your timeline and budget. Family C is a preference rather than a requirement for most authors, and it is the first thing to relax when results thin out.
| Family | Question it answers | Typical selectivity | Relax when… |
|---|---|---|---|
| A · Scope & standing | Will this journal count? | High; cuts the corpus sharply | Last. This is usually the requirement. |
| B · Process & policy | Can I work with this journal? | Moderate | Second. Widen the APC ceiling before dropping the review model. |
| C · Provenance | Do I want this journal? | Variable; often severe in combination | First. Country and publisher narrow fastest. |
Table 5.1Relaxation order. Read in reverse, it is also the order in which to add filters when a result set is too wide.
Each of §6 to §14 has the same six elements: what the indicator controls; its cue words; the values it accepts; worked examples marked ✓; failure modes marked ✕; and a closing note where there is a subtlety worth knowing. Read the failure modes even if you skip the rest, because they explain most unexpected results. Select Reading beside any example to see how the workbench decomposes it.
§6·Indicator 1·Family A · Scope & standing
Restricts results to journals covered by a named abstracting and indexing service.
Cue words
Any of these forms activates the filter. The short forms (index, indexed, indexing) are accepted so that natural phrasings such as Scopus-indexed journals resolve correctly. When no service name follows a short form, as in journals indexing palm oil research, the phrase stays in the research input.
Accepted values
The name of any service in the platform’s coverage list of 76 indexing databases. Give names in their common short form (Scopus, ERA, MyCite, Scimago, DOAJ); matching is case-insensitive. A representative selection appears in Appendix A.
Several services may be named in one clause. They combine disjunctively: a journal qualifies if any of them covers it.
Worked examples
Two services, one clause. Returns journals covered by either.
A regional index. Useful where a national research assessment recognises it.
Combines with Indicator 2; see §16 on stacking.
Failure modes
No cue word. In alone is not a trigger, so this reads as a research topic about coverage and no filter is applied.
A reputable database is not a name. The clause resolves to nothing and is discarded.
A misspelt service name matches no journal and can empty a result set on its own (§18, step 5).
Inclusion in an index means a service has judged a journal to meet its selection criteria and has agreed to abstract its content. It does not warrant editorial rigour, and coverage is revocable: services routinely discontinue titles that cease to meet their criteria. A journal may therefore advertise an index it no longer appears in. Where the distinction matters for a submission decision, pass the shortlist through Journal Risk and Quality Assessment (§1) rather than relying on the indexing filter alone.
Curated citation indexes such as Scopus and the Web of Science Core Collection admit journals after editorial evaluation. Discovery engines and metadata registries such as Google Scholar, Crossref or BASE admit content largely automatically, so coverage there establishes discoverability rather than evaluation. Two names need care. Scimago denotes the SCImago Journal & Country Rank portal, which ranks journals on citation data from Scopus; it is a ranking source, not an independent index (SCImago, n.d.). PubMed is a search interface over several collections, and a journal found in PubMed is not necessarily indexed in MEDLINE, its selective component (National Library of Medicine, n.d.). Appendix A groups the services accordingly.
§7·Indicator 2·Family A · Scope & standing
Restricts results by the journal’s position in the ranked distribution of its subject category.
Cue words
Uniquely among the nine indicators, the value is the cue word. No introducing phrase is required, although ranked, rated or in may be used for readability.
Accepted values
Any single quartile, or any combination. Combinations are disjunctive and need not be contiguous.
| You write | Interpreted as |
|---|---|
| Q1 | First quartile only: the top 25 % of the category |
| Q1, Q2 | First or second quartile: the upper half |
| Q3, Q4 | Third or fourth quartile: the lower half |
| Q1, Q4 | Either extreme; non-contiguous combinations are permitted |
Worked examples
The canonical pairing: the index establishes the ranking source, the quartile sets the threshold.
Excludes only the bottom quartile: a pragmatic setting for early-career submissions.
Failure modes
Not a quartile expression. Write Q1 or Q1, Q2 explicitly.
Quartiles take no comparative operators. Enumerate instead: Q1, Q2.
A quartile is not a property of a journal in isolation. It is the journal’s rank position within a specific subject category, computed from a citation metric (SJR in the SCImago system, the Journal Impact Factor in Journal Citation Reports) and recomputed annually. A title assigned to three categories may be Q1 in one, Q2 in another and Q3 in a third, and each statement is correct.
Two consequences follow for query design. First, combining a quartile with a subject area (Indicator 7) produces a materially different and usually more meaningful result than a quartile alone, because it fixes the category in which the rank is read. Second, check a quartile claim on a journal’s own website against the category it refers to; publishers tend to advertise the strongest one.
SJR quartiles and Journal Impact Factor quartiles are computed on different citation databases, category schemes and metrics, so one journal can hold different quartiles in each. SJR weights each citation by the prestige of the citing journal (González-Pereira et al., 2010); the Journal Impact Factor divides one year’s citations to a journal’s items from the two preceding years by the number of citable items it published in those years (Garfield, 2006). Naming the index alongside the quartile, as in the canonical pairing above, fixes which ranking you mean.
§8·Indicator 3·Family B · Process & policy
Restricts results by how often the journal issues content.
Cue words
As with quartiles, the value is the cue. Publishing, issued or appearing may precede it for readability.
Accepted values
| Value | Issues per year | Practical implication |
|---|---|---|
| Monthly | 12 | Shortest queue to appearance once accepted |
| Bi-monthly | 6 | Common in established society journals |
| Quarterly | 4 | The most frequent setting across the corpus |
| Tri-annual | 3 | Typical of regional and institutional titles |
| Semi-annual | 2 | Longer wait between acceptance and issue |
| Annual | 1 | Yearbooks, proceedings series |
| Continuous publication | — | Articles appear on acceptance; no issue queue |
| Article-by-article | — | As above; publishers use the two terms interchangeably |
| Irregular | — | No fixed schedule; timing cannot be predicted |
Worked examples
The pattern used on the platform’s own example card.
Appropriate where time to appearance is the binding constraint.
Failure modes
Not a listed value. Name the interval.
Frequency takes no comparative operators. Enumerate: publishing monthly, bi-monthly or quarterly.
Issue frequency governs only the final step: the wait between acceptance and appearance. It says nothing about how long review takes, which is the larger share of total time to publication for most submissions. A quarterly journal with a six-week review cycle will publish you sooner than a monthly journal with a nine-month one. Where speed is the objective, prefer Continuous publication or Article-by-article, which remove the issue queue entirely.
Bi-monthly is read as six issues a year. In general English the word can also mean twice a month, and no value is listed for that schedule. Because frequency values are self-cueing (§15), a research topic that happens to contain one, such as annual rainfall variability, risks acquiring a frequency filter. If results look unexpectedly narrow, rephrase the topic (year-to-year rainfall variability).
§9·Indicator 4·Family B · Process & policy
Restricts results by the review procedure the journal operates.
Cue words
Accepted values
| Model | Who is concealed | Character |
|---|---|---|
| Single blind | Reviewers, from authors | The long-standing default |
| Double blind | Authors and reviewers, from each other | The most requested model |
| Triple blind peer review | Both parties, plus authorship from the editor | The most stringent model |
| Open peer review | Nobody; reports are often published | Reports join the public record |
| Post-publication peer review | — | Review after appearance |
| Editorial review | — | In-house; no external referees |
| Not blinded | Nobody | Identities known throughout |
| Peer review | — | Matches any refereed model |
Worked examples
The most frequently used clause for this indicator.
Narrow: comparatively few journals operate this model.
Selects journals that publish referee reports alongside the article.
The bare phrase blind peer review is a listed value, but it does not distinguish single from double concealment, and journals use it inconsistently in their own documentation. Write single blind or double blind explicitly.
Double-blind review conceals identity; it does not guarantee anonymity. In specialised fields, self-citation patterns, distinctive methods, named datasets and preprint deposits often make authorship recoverable by a well-read referee. Filtering on the model is reasonable; treating it as decisive is not.
The peer-review terminology standard ANSI/NISO Z39.106-2023 recommends anonymized in place of blind (single-anonymized, double-anonymized), and many journals now describe their process that way (NISO, 2023). The Journals Assistant’s cue words use blind, so write double blind even when a journal’s own page says double-anonymized.
Open peer review also needs care. A systematic review found 122 definitions of the term, resolving into 22 distinct configurations of seven traits, including open identities, open reports and open participation (Ross-Hellauer, 2017). Check which of these a matched journal actually practises.
§10·Indicator 5·Family C · Provenance
Restricts results to journals published in a named country. The corpus covers more than 150.
Cue words
Put the cue immediately before the country name. The specification also lists bare in, but it is the weakest cue in the set: in the matched pairs of §15, Health policy in Malaysia extracts nothing. Treat bare in as unreliable and write published in, from or country.
Accepted values
Any country name in English. Demonyms, colloquial forms and common abbreviations are normalised automatically to the standard English name in Title Case; the full mapping is in §20. A representative extract:
| You may write | Normalised to |
|---|---|
| iranian | Iran |
| american · USA · US | United States |
| british · UK | United Kingdom |
| uae | United Arab Emirates |
| korean | South Korea |
| dutch | Netherlands |
Worked examples
The explicit form. Unambiguous, and the safest phrasing.
Natural prose; the cue phrase carries the instruction.
Failure modes
Malaysia is part of your research topic. It is treated as subject matter, not as a filter, and correctly so.
A country word inside a journal title is never extracted. Canadian will not become a country filter.
Scholarly publishing is geographically concentrated: a large share of high-quartile titles originate in a small number of countries. Combining country with Q1 therefore empties the result set more often than any other pairing in this guide. If you need both, consider whether a regional index (Indicator 1) expresses your intent better than a national constraint. ASEAN Citation Index or MyCite, for instance, capture regional standing without fixing a single country.
§11·Indicator 6·Family C · Provenance
Restricts results to journals issued by a named publishing house, society or press.
Cue words
From is shared with Indicator 5, and the value decides which indicator it serves: from Indonesia sets a country, from Elsevier a publisher.
Accepted values
The publisher’s name as you would say it, trimmed of surrounding words and cast to Title Case. Supply the name only. The system neither expects nor uses a full institutional address, legal suffix or imprint hierarchy.
Illustrative only. Any publisher present in the corpus may be named.
Worked examples
The explicit form.
Natural prose form; equally reliable.
Failure modes
Frontiers is inside a journal title and is deliberately not extracted as a publisher. This is the canonical illustration of the cue-signal principle (§15).
Over-specified. The address is discarded and may interfere with matching. Write publisher Springer.
Filtering by publisher is useful in three situations: when your institution holds a read-and-publish or transformative agreement that waives article processing charges with a particular house; when you are targeting a society journal whose readership is your intended audience; and when a funder mandate restricts you to outlets with specific licensing terms. Outside these cases it is a blunt instrument. Large publishers carry titles of every quartile and every subject, so the filter narrows provenance without improving fit. Prefer Indicators 1, 2 and 7 for that.
§12·Indicator 7·Family A · Scope & standing
Restricts results to journals classified under a named disciplinary category.
Cue words
This is the indicator most often misapplied, because its values look exactly like research topics. A subject is extracted only when a cue word marks it as a filter. A discipline word that is simply what your paper is about, with no cue, stays in the research input, where it belongs. Education technology adoption is a topic. Education technology adoption, subject Education is a topic plus a filter.
Accepted values
Any category in the platform’s controlled vocabulary, written in Title Case. Your own terminology is accepted where it falls outside the list. Values attested in the corpus include:
The corpus holds many more. See Appendix B.
Worked examples
Topic and filter clearly separated. Returns agriculture journals that publish on palm oil.
topic: works as a subject cue despite the word, a common source of confusion. Keep the colon.
Failure modes
Surgical is the topic, not a filter. No subject constraint is applied, which is the correct behaviour.
A subject word inside a journal title is never extracted.
Because quartile rank is computed within a subject category (§7), naming a subject alongside a quartile fixes the category in which the rank is read and materially sharpens the result. Q1 alone asks for any journal that is first-quartile in any of its categories; Q1, subject Food Science asks for journals that are first-quartile in food science. The second is almost always what the author means.
The worked examples in §4 show that a topic followed by journals (Halal tourism journals) stays in the research input. The implicit form <subject> journals can therefore apply only when the preceding words name a category, as in Food Science journals. Write subject Food Science when you want certainty. The cue index lists topic: with its colon; without it, topic is an ordinary word that may belong to your research input.
§13·Indicator 8·Family C · Provenance
Restricts results to journals that publish in a named language.
Cue words
Accepted values
Language names in English, Title Case. Colloquial names are normalised automatically. Multiple languages is itself a value, matching journals that accept submissions in more than one.
| You may write | Normalised to |
|---|---|
| farsi | Persian |
| mandarin | Chinese |
Worked examples
Explicit form. Returns Indonesian-language journals publishing on palm oil.
Prose form; equally reliable.
Failure modes
Language appears, but as part of the research topic. No filter is applied, correctly.
A language word inside a journal title is never extracted.
Many journals publish in a national language while carrying English abstracts, keywords and metadata, and it is the metadata that most indexing services abstract. Filtering to a non-English language therefore selects for regional readership, not for reduced discoverability. Conversely, filtering to English does not exclude journals based in non-anglophone countries: a large proportion of titles worldwide publish in English regardless of provenance, so this filter combines poorly with Indicator 5 and rarely adds information alongside it.
§14·Indicator 9·Family B · Process & policy
Restricts results by the fee charged to authors on acceptance. The only indicator that accepts numeric ranges.
Cue words
Values: plain numbers only
Express the amount as a bare number, never with a currency symbol or code. The corpus stores amounts as 2200 USD; your query supplies only 2200.
| You write | Interpreted as | Reading |
|---|---|---|
| APC under 1000 | max: 999 | Strictly below |
| APC at most 2000 | max: 2000 | Inclusive ceiling |
| APC between 500 and 1500 | min: 500 · max: 1500 | Inclusive band |
| APC of exactly 300 | eq: 300 | Exact match |
Table 9.1Under excludes the boundary; at most includes it. Generalised in §17.
No-fee journals
All of the following resolve to eq = 0: journals that levy no author charge at all.
The phrases open access and gold OA are not treated as fee constraints, and no APC filter is set when they appear alone. Gold open access describes the licence and availability of the article, not who pays for it, and under that model the author or their funder typically pays. Only the diamond variant implies no author fee. Diamond is sometimes called platinum, but the listed cue is diamond.
Worked examples
Returns journals charging 999 or less.
Resolves to eq = 0: no author charge.
Failure modes
Currency symbol, code and thousands separator. Write the bare number: APC under 1000.
Inside a pasted abstract. A cost figure in your research input is never read as a constraint on the journal.
Set the ceiling from what is actually available: a grant line, a publishing fund, a transformative agreement. Many publishers operate waiver schemes that the listed charge does not reflect, so a journal slightly above your ceiling may still be within reach.
Diamond journals are numerous (the OA Diamond Journals Study estimated up to 29,000 of them) but predominantly small, often reliant on volunteer labour, and about 60 % of them are in the humanities and social sciences (Bosman et al., 2021). That profile helps explain why free combined with Q1 often returns nothing outside those fields (Table 16.1).
Part III
The cross-cutting rules: how the extractor separates instruction from subject matter, how filters combine, how numeric constraints are expressed, and what to do when a query returns nothing at all.
§15
One rule, stated five times in the source specification. It explains most surprising results.
Country, publisher, subject and language share a structural problem, and fee amounts share a version of it: their values are ordinary words or figures that occur naturally inside journal titles, research topics and pasted abstracts. Canadian, Frontiers, Surgery and Spanish are respectively a country, a publisher, a subject and a language, and they are also ordinary parts of titles and topics. If the extractor treated every occurrence as a filter, almost every query would be silently over-constrained.
A value becomes a filter only when a cue word marks it as one. The same word, unmarked, remains part of the research input.
Two corollaries follow, and both are enforced: a value appearing inside a journal title is never extracted, and a value that is simply what your paper is about, with no cue, is never extracted.
| Query | Extracted as filter | Why | Try |
|---|---|---|---|
| Health policy in Malaysia | Nothing | Malaysia is the subject of the research. Correct: you want papers about Malaysian health policy. | |
| Health policy, country Malaysia | Country = Malaysia | The cue word country marks it as an instruction. | |
| Frontiers in Bioengineering | Nothing | Frontiers sits inside a journal title. | |
| Bioengineering, published by Frontiers | Publisher = Frontiers | The cue phrase published by marks it. | |
| Language acquisition in bilingual children | Nothing | Language belongs to the topic. | |
| Bilingual children, written in Spanish | Language = Spanish | The cue phrase written in marks it. | |
| Surgical site infection rates | Nothing | Surgical is the topic. | |
| Infection rates, subject Surgery | Subject = Surgery | The cue word subject marks it. |
Table 15.1Matched pairs. In each pair the same word appears twice; only the cued instance becomes a filter.
Six of the nine indicators require a cue. For five of them the requirement is critical, because their values are ordinary words. For the other three the value is its own cue.
| Indicator | Cue required | Note |
|---|---|---|
| 1 · Indexing database | Yes | Database names are distinctive, but a cue is still required |
| 2 · Quartile | Self-cueing | “Q1” has no other meaning in this context |
| 3 · Frequency | Self-cueing | “Quarterly” has no other meaning in this context; see the caution in §8 for words such as annual |
| 4 · Peer-review model | Self-cueing | Review-model names are unambiguous |
| 5 · Country | Yes, critical | Country names occur constantly in titles and topics |
| 6 · Publisher | Yes, critical | Publisher names occur inside journal titles |
| 7 · Subject area | Yes, critical | Subject words are usually also the research topic |
| 8 · Language | Yes, critical | Language words occur in titles and in linguistics topics |
| 9 · APC | Yes, critical | Monetary figures occur inside pasted abstracts |
It is tempting to read this as a limitation, the system failing to notice something obvious. The opposite is true. Without it, pasting an abstract that mentions a cost figure would silently impose a fee ceiling; a query about Canadian health policy would silently restrict itself to Canadian journals; and any mention of a discipline would collapse the search onto that single category. The restriction is what makes it safe to paste a full manuscript into the same box you use to write instructions. The price is that instructions must be marked, and this guide is largely an inventory of the available marks.
§16
Filters combine conjunctively across indicators and disjunctively within one. Knowing which is which is the difference between narrowing a search and emptying it.
Across indicators, filters combine with logical AND: a journal must satisfy every indicator you name. Within an indicator, listed values combine with logical OR: a journal need satisfy only one of them.
Here is the fully specified query from §4 in logical form:
The asymmetry has a practical consequence. Adding a value to an existing indicator always widens the result set. Adding a new indicator always narrows it. When a query returns too little, first enumerate additional values inside the indicators you already have (Q1 becomes Q1, Q2) before you remove an indicator entirely.
| Pairing | Behaviour | Comment |
|---|---|---|
| Indexing × Quartile | Strong | The canonical pairing. The index supplies the ranking source; the quartile supplies the threshold. |
| Quartile × Subject | Strong | Fixes the category in which rank is computed. Almost always what the author intends. |
| Subject × Indexing | Strong | Discipline plus recognition: a well-formed institutional requirement. |
| Review model × APC | Neutral | Independent properties; they combine without interference. |
| Country × Language | Redundant | Substantially correlated. The second filter rarely adds information and may exclude English-language titles from the country you named. |
| Country × Q1 | Fragile | High-quartile titles concentrate in few countries. The most common cause of an empty result. |
| Publisher × Country | Fragile | Large publishers are registered in a small number of jurisdictions; the pair often contradicts itself. |
| Free APC × Q1 | Fragile | Diamond open access is uncommon at the top of most categories. Try under 500 instead of free. |
Table 16.1Interaction profile of common pairings. Fragile means the pair is logically valid but frequently satisfied by nothing in the corpus. The workbench flags these pairings as you type.
There is no technical limit on the number of filters. There is a practical one, and it varies with the richness of your research input (§2). As a rule of thumb, a one-line topic tolerates five or six indicators, an abstract three or four, and a full manuscript two or three. Beyond that the constraint set tends to be empty, and the query stops being a search and becomes an assertion about what ought to exist.
§17
One indicator, the article processing charge, accepts quantities. This section documents the grammar in full, including the boundary conventions.
A numeric constraint resolves to a triple of minimum, maximum and exact value, any of which may be unset. The phrasing you choose determines which of the three are populated and whether the boundary you name is included or excluded. Amounts are whole numbers, so a strict bound moves by one: under 1000 becomes a maximum of 999, and over 1000 a minimum of 1001.
| Phrasing | min | max | eq | Boundary |
|---|---|---|---|---|
| under N · below N · less than N | — | N−1 | — | Excluded: N itself does not qualify |
| at most N · up to N · no more than N | — | N | — | Included |
| over N · above N · more than N | N+1 | — | — | Excluded |
| at least N · from N | N | — | — | Included |
| between A and B | A | B | — | Both included |
| exactly N · of N | — | — | N | Exact match only |
| free · no APC · diamond OA | — | — | 0 | No author charge of any kind |
Table 17.1The complete numeric grammar. The under / at most distinction is the one users most often get wrong: under 1000 returns nothing priced at exactly 1000.
Bare integer. The only correct form.
Currency symbol.
Currency code. The corpus stores 1000 USD; your query supplies only the number.
Thousands separator. Write 1000.
If you paste an abstract containing “the mean cost per episode was 1,240”, no fee filter is created. The numeric grammar applies only to an explicit constraint on the journal, introduced by one of the Indicator 9 cue words. This is the cue-signal principle (§15) applied to quantities.
These are materially different requests. free resolves to eq = 0 and admits only journals with no author charge whatsoever: the diamond open-access model, in which a society, consortium or institution underwrites publication. under 500 admits those same journals and every journal charging a nominal fee below that ceiling, which in many fields is a substantially larger and more useful set. Unless zero cost is an absolute constraint, prefer the ceiling.
§18
A disciplined procedure for both failure directions, and the order in which to intervene.
An empty return from a well-formed query is a statement about the corpus, not a malfunction. Work through the following in order, changing one thing at a time. Changing several at once tells you nothing about which constraint was binding.
Drop country, then publisher, then language. These are preferences for most authors, and they narrow fastest. If results appear, you have found the binding constraint.
Extend Q1 to Q1, Q2; extend a single index to two; raise an APC ceiling. Because values combine disjunctively (§16), this widens without abandoning the requirement.
Diamond open access is scarce in most categories. Under 500 is a far larger set than free and usually meets the same practical need.
Only now. If you supplied a full manuscript, try the abstract; if the abstract, try a topic line. A narrower input means a wider semantic match, which gives the filters more to act on.
Confirm that every intended filter carries a cue from §19. A missing cue widens rather than narrows, so this step matters more for the opposite failure, but a misspelt value (Scoups) resolves to nothing and can empty a set on its own.
The workbench applies this procedure to whatever query it holds, and offers each step as a single change.
The reverse failure is more insidious, because it produces a plausible-looking list rather than an obvious error. Almost always, a filter you believed you had applied was never extracted.
These procedures are diagnostic, not cosmetic. Their value lies in isolating which constraint is binding, and that isolation is lost the moment you alter two things at once. Treat successive queries as a series, each differing from the last in exactly one respect, and the corpus will show you precisely where its limits lie.
Part IV
Lookup material: every cue word in one index, the normalisation tables, a symptom-driven troubleshooting matrix, a glossary, and appendices of controlled vocabulary, query templates and sources.
§19
Every trigger phrase in the system, alphabetically, with the indicator it activates and the shape of the value it expects.
| Cue word or phrase | Indicator | Expected value |
|---|---|---|
| Annual | 3 · Frequency | Self-valuing |
| APC | 9 · APC | Bare number, or a no-fee phrase |
| area: | 7 · Subject area | Category name, Title Case |
| article processing charge | 9 · APC | Bare number |
| Article-by-article | 3 · Frequency | Self-valuing |
| Bi-monthly | 3 · Frequency | Self-valuing; six issues a year |
| Blind peer review | 4 · Review model | Self-valuing, but qualify it (§9) |
| Continuous publication | 3 · Frequency | Self-valuing |
| cost to publish | 9 · APC | Bare number |
| country | 5 · Country | English country name |
| diamond OA · diamond open access | 9 · APC | Resolves to eq = 0 |
| domain: | 7 · Subject area | Category name, Title Case |
| Double blind · Double blind peer review | 4 · Review model | Self-valuing |
| Editorial review | 4 · Review model | Self-valuing |
| fee · publication fee · publishing fee | 9 · APC | Bare number |
| fee-free · free · no-cost publishing | 9 · APC | Resolves to eq = 0 |
| from <country> | 5 · Country | English country name |
| from <publisher> | 6 · Publisher | Publisher name, Title Case |
| in the field of | 7 · Subject area | Category name, Title Case |
| index · indexes · indexed · indexing | 1 · Indexing database | Service name; several permitted |
| indexed by · indexed in · indexed with | 1 · Indexing database | Service name; several permitted |
| indexing in · listed in · covered by | 1 · Indexing database | Service name; several permitted |
| Irregular | 3 · Frequency | Self-valuing |
| language | 8 · Language | Language name in English |
| Monthly | 3 · Frequency | Self-valuing |
| no APC · no fee · no publication fee · no article processing charge · zero APC · waived fees | 9 · APC | Resolves to eq = 0 |
| Not blinded | 4 · Review model | Self-valuing |
| Open peer review | 4 · Review model | Self-valuing |
| Peer review | 4 · Review model | Generic; matches any refereed model |
| Post-publication peer review | 4 · Review model | Self-valuing |
| processing charge | 9 · APC | Bare number |
| published by · publisher | 6 · Publisher | Publisher name, Title Case |
| published in <country> | 5 · Country | English country name |
| Q1 · Q2 · Q3 · Q4 | 2 · Quartile | Self-valuing; combinations permitted |
| Quarterly | 3 · Frequency | Self-valuing |
| Semi-annual | 3 · Frequency | Self-valuing |
| Single blind | 4 · Review model | Self-valuing |
| subject · subject: · <subject> journals | 7 · Subject area | Category name, Title Case |
| topic: | 7 · Subject area | Category name. Note that this is a subject cue |
| Tri-annual | 3 · Frequency | Self-valuing |
| Triple blind peer review | 4 · Review model | Self-valuing |
| written in | 8 · Language | Language name in English |
| No cue words match. Try a shorter term. | ||
Table 19.1Self-valuing means the cue word is the value; no separate term follows it. Bare in is deliberately absent: it is not a reliable country cue (§10).
§20
How informal input is converted to the standard forms held in the corpus. You may write either column; the system stores the right-hand one.
| You may write | Stored as | You may write | Stored as |
|---|---|---|---|
| american · USA · US | United States | indonesian | Indonesia |
| brazilian | Brazil | iranian | Iran |
| british · UK | United Kingdom | japanese | Japan |
| chinese | China | korean | South Korea |
| dutch | Netherlands | polish | Poland |
| german | Germany | saudi | Saudi Arabia |
| indian | India | uae | United Arab Emirates |
Table 20.1The mapping is not limited to these entries: any country is accepted and returned in standard English Title Case. The corpus covers more than 150. The rows above are the cases where informal usage diverges most from the stored form.
| You may write | Stored as |
|---|---|
| farsi | Persian |
| mandarin | Chinese |
| Indicator | Rule applied on extraction |
|---|---|
| Country | Standard English name, Title Case. Demonyms and codes are resolved first. |
| Publisher | Trimmed and Title Cased, exactly as you typed it. No institutional address is inferred or added. |
| Subject area | Title Case. Your own terminology is preserved where it falls outside the controlled vocabulary. |
| Language | English name, Title Case. Colloquial names are resolved first. |
| APC | Reduced to a plain integer. The query must supply the bare number; the corpus reconciles the unit. |
§21
Symptom, cause, remedy. Ordered by how often each arises in practice.
| Symptom | Probable cause | Remedy |
|---|---|---|
| A filter appears to have been ignored | The value was written without a cue word and stayed in the research input (§15) | Add the explicit cue from §19 |
| No results at all | Over-constraint. Most often country × quartile, or free × Q1 | Relax Family C first, then widen within indicators (§18) |
| Wrong discipline dominates | No subject filter, and a short research input | Add subject <area>, or supply an abstract instead of a topic line |
| Quartiles look inconsistent | Quartile is category-relative; without a subject, any category qualifies (§7) | Pair the quartile with a subject area |
| Fee-bearing journals returned after asking for open access | Open access and gold OA set no fee constraint (§14) | Write free, no APC, diamond OA, or a numeric ceiling |
| A journal priced at exactly the ceiling is missing | under N excludes N; at most N includes it (§17) | Use at most where the boundary should qualify |
| APC filter has no effect | A currency symbol, code or thousands separator was included | Write the bare integer: under 1000 |
| Publisher filter returns nothing | A full legal name or address was supplied | Give the short name only: publisher Springer |
| Results ignore an instruction inside an attached file | Filters embedded in an uploaded document are read as content (§2) | Type preference clauses into the question box alongside the attachment |
| A manuscript query returns almost nothing | Strong content signal combined with heavy filtering (§2) | Reduce to two or three indicators, or submit the abstract instead |
| Preference clauses appear to do nothing at all | A different assistant is selected in the model dropdown (§1) | Switch to Journals Assistant and resubmit |
§22
Terms used in this guide, defined as the platform uses them.
eq = 0.No terms match.
Appendix A
Services you may name in an Indicator 1 clause, grouped by what inclusion in each one signals.
The platform’s authoritative coverage list runs to 76 indexing databases and is maintained within the application, where it is updated as services are added or discontinued. That list was not part of the source specification from which this guide was prepared. What follows is a representative selection of widely recognised services, offered to show the range of what may be named, not a reproduction of the platform’s list. Consult the application before relying on a specific entry.
Journals are admitted after editorial evaluation. SCIE, SSCI, AHCI and ESCI are journal indexes within the Web of Science Core Collection.
SCImago Journal & Country Rank ranks journals on Scopus citation data and publishes the SJR quartiles (§6, §7).
Coverage is largely automatic. It establishes that a journal’s content is discoverable, not that it has been evaluated.
ERA: Excellence in Research for Australia. The ERA 2023 evaluation round was not held, and the ERA 2023 Submission Journal List remains the most recent edition of the list (Australian Research Council, 2022). MyCite: Malaysian Citation Index.
| Field | Services | Note |
|---|---|---|
| Medicine & life sciences | PubMed · MEDLINE · PubMed Central · Embase · CINAHL · BIOSIS · Zoological Record | MEDLINE is the selective component of PubMed; PubMed Central is a full-text archive |
| Psychology & behaviour | PsycINFO · SPORTDiscus | |
| Engineering & computing | IEEE Xplore · Inspec · Compendex | IEEE Xplore is a publisher platform; Inspec and Compendex are independent indexes |
| Chemistry | CAS (Chemical Abstracts Service) | |
| Mathematics | MathSciNet · zbMATH | |
| Agriculture & food | AGRICOLA · CAB Abstracts · PubAg | |
| Earth sciences | GeoRef | |
| Astronomy & physics | Astrophysics Data System · arXivpreprint server | |
| Education | ERIC | |
| Economics & social science | EconLit · RePEc · SSRNpreprint server | |
| Transport | TRID |
DOAJ is a curated directory of open-access journals. EBSCOhost and ProQuest are aggregator platforms, JSTOR is a digital archive, and Ulrich’s is a serials directory.
Use the common short form: Scopus, not Elsevier’s Scopus abstract and citation database. Matching is case-insensitive. Several services may be named in a single clause and combine disjunctively: indexed in ERA, Scopus returns journals covered by either.
Appendix B
Values attested in the corpus for Indicator 7. Your own terminology is accepted where a category is not listed.
| Grouping | Attested values |
|---|---|
| Health & clinical | Medicine · Clinical Medicine · Surgery · Ophthalmology · Immunology |
| Life sciences | Biochemistry · Genetics and Molecular Biology · Microbiology · Biotechnology |
| Applied & agricultural | Food Science · Chemical Technology · Agriculture |
| Physical & formal | Science · Engineering · Mathematics |
| Social & humanistic | Social Sciences · Education · Law · Arts and Humanities |
| Cross-cutting | Multidisciplinary |
Table B.1Groupings are editorial aids for this guide; the corpus holds the values as a flat vocabulary, and many more than are listed here.
Supply your own term in Title Case behind a subject cue: subject Veterinary Science, in the field of Public Health. The value is preserved as written. If a query with an unlisted subject returns nothing, try the nearest broader category from the table above before concluding that no such journals exist.
Appendix C
Nine patterns covering common submission scenarios. Fill in the placeholders and submit.
| Scenario | Template | Use |
|---|---|---|
| First orientation | [topic] | |
| Standard submission search | [topic] journals indexed in [service], Q1, Q2, offering double blind peer review | |
| Institutional requirement | [abstract] … indexed in [service], ranked Q1, subject [area] | |
| Budget-constrained | [topic] journals indexed in [service], Q1, Q2, publishing fee under [amount] | |
| No funds available | [topic] journals indexed in DOAJ, diamond open access, subject [area] | |
| Speed is the priority | [abstract] … indexed in [service], continuous publication, Q1, Q2 | |
| Regional visibility | [topic] journals indexed by [regional index], published in [country], written in [language] | |
| Publisher agreement | [abstract] … published by [publisher], Q1, Q2, subject [area] | |
| After a desk rejection | [full manuscript] … indexed in [service], Q2, Q3, double blind peer review |
Adapting a template.Each template is deliberately under-constrained, leaving room for the condition that matters most in your case. Add filters in the order given in Table 5.1 (Family A, then B, then C) and remove them in reverse if results thin out.
Appendix D Web edition
Journal-level indicators describe venues, not articles. This appendix sets the filters in this guide against the principles that now govern research assessment, and lists the sources cited in the editorial notes.
The indicators in Part II help you find a journal that fits your manuscript, your timeline and your institution’s requirements. They are not measures of the quality of your work, and assessment frameworks increasingly say so. The San Francisco Declaration on Research Assessment asks that journal-based metrics such as the Journal Impact Factor not be used as a surrogate for the quality of individual articles (DORA, 2012). The Leiden Manifesto sets out ten principles for research metrics, the first being that quantitative evaluation should support qualitative, expert assessment (Hicks et al., 2015). The Agreement on Reforming Research Assessment commits its signatories to assessment based primarily on qualitative judgement, supported by responsible use of quantitative indicators (CoARA, 2022).
Screening matters as much as matching. A consensus definition describes predatory journals and publishers as entities that prioritise self-interest at the expense of scholarship, characterised by false or misleading information, deviation from best editorial and publication practices, a lack of transparency, and aggressive and indiscriminate solicitation (Grudniewicz et al., 2019). No preference filter tests for these properties. Before you commit a manuscript, use the Journal Risk and Quality Assessment card (§1) and an independent checklist such as Think. Check. Submit.
Colophon
Prepared from the i-Future AI User Preference for Journal Assistant AI specification of 24 September 2026, incorporating the correction schedule of 26 September 2026. The nine indicators, their cue words and their accepted values are documented as specified. The three-family taxonomy (§5), the interaction profile (Table 16.1), the diagnostic procedures (§18), the troubleshooting matrix (§21), the glossary and the query templates are editorial additions, written to make the specification usable as a working reference.
This web edition adds the query workbench, the boundary explorer, the template builder, the notes marked Editorial note, and Appendix D.
Where this guide and the application disagree, the application is authoritative. This applies in particular to the coverage list discussed in Appendix A and the subject vocabulary in Appendix B, both of which are maintained in-product and change over time.
The following points in the print edition 2.1 disagreed with one another. Each has been resolved as described, pending confirmation against the specification.