Web of Science Advanced Search: TS=, AU=, Boolean

To search Web of Science effectively, use Advanced Search with field tags (TS=, AU=, SO=), Boolean and proximity operators, and Cited Reference Search rather than typing a plain-text query into Basic Search, which returns thousands of loosely related records. Author-disambiguation tools like ResearcherID and ORCID keep an author search from missing or merging the wrong person.

For what Web of Science itself is — its indexes, ownership history, and how it differs from Scopus and OpenAlex — see CASRAI’s Web of Science dictionary entry and the Scopus vs Web of Science vs OpenAlex comparison. This page assumes you already know what the platform is and focuses only on searching it well.

Basic Search vs. Advanced Search

Basic Search is Web of Science’s default entry point: a single search box (defaulting to a Topic search across title, abstract, author keywords, and Keywords Plus) with a small number of drop-down fields — Author, Title, Publication Name, Year — that can be chained together with AND/OR using “+ Add row.” It is fast for a quick lookup (“find this paper,” “what has this author published recently”) but its field options are limited to what the interface exposes, and complex logic (nested Boolean groups, proximity operators, exclusions with NOT) is awkward or unavailable.

Advanced Search uses a single free-text query box in which you type field tags and operators directly, e.g. TS=("data management plan*") AND AU=(Smith J). It is the only interface that supports:

The full set of field tags (not just the handful exposed as Basic Search dropdowns)
The NOT operator
Nested parenthetical logic combining multiple fields and operators in one query
Proximity operators (NEAR/x, SAME)
Saved, re-runnable, and exportable search strings — useful for documenting a systematic-review search strategy or re-running a saved alert

For anything beyond a one-off lookup — a literature review, a systematic search, a recurring current-awareness search, or a search you need to defend or reproduce later — build it in Advanced Search rather than the Basic Search dropdowns.

Field Tags: Searching a Specific Part of the Record

A field tag tells Advanced Search which part of the record to search, instead of matching your terms anywhere in a record. Tags are written as a two-letter code followed by an equals sign and the search term in parentheses. The tags most relevant to a practical literature search:

TS= (Topic) — searches title, abstract, author keywords, and Keywords Plus together in one field. This is the closest Advanced Search equivalent to Basic Search’s default box, and the field most searches should start from.
TI= (Title) — searches only the article title. Narrower than TS=; useful when a topic search returns too much noise from abstract-only mentions.
AB= (Abstract) — searches only the abstract field.
AU= (Author) — searches the author name field. Because author names are ambiguous (see the disambiguation section below), Web of Science also supports author-identifier fields for a more precise version of the same search.
AI= (Author Identifiers) — searches by a Web of Science ResearcherID or a linked ORCID iD rather than a typed name, which sidesteps name-variant and common-name ambiguity entirely.
AD= (Address) — searches institution and place names in the Address field. Note Clarivate’s own caveat: it searches the address field “usually for the first named author of an article,” so an address search is not a reliable way to retrieve every paper with any author at a given institution. For affiliation work, prefer the organisation tags below.
SO= (Publication Name / Source) — restricts results to a specific journal or source title, e.g. SO=(Nature).
PY= (Year Published) — restricts by publication year or year range.
SU= (Research Area) / WC= (Web of Science Category) — restricts by the subject classification Web of Science assigns to a journal.
DO= (DOI) — looks up a specific record directly by its DOI, useful for confirming a record exists in Web of Science’s coverage rather than a keyword search.
AK= / KP= (Author Keywords / Keywords Plus) — search the two keyword fields individually rather than as part of a combined Topic search, when you specifically want author-supplied keywords versus the algorithmically-derived Keywords Plus terms.
ALL= (All Fields) — searches every indexed term at once. Convenient, but it carries three documented restrictions the other tags do not: Funding Details are excluded from an All Fields search, it does not support the SAME or NEAR proximity operators, and only right-hand truncation works in it (oxid* is valid, *oxide returns an error).
UT= (Accession Number) and PMID= — direct record lookups by Web of Science accession number or PubMed ID.
IS= — searches ISSN and ISBN; Clarivate documents both under the same tag.
GP= (Group Author) and ED= (Editor) — for corporate/consortium authorship and for edited volumes, both of which a plain AU= search will miss.
FO= / FG= / FT= (Funding Agency / Grant Number / Funding Text) — the funding acknowledgement fields. Directly useful to a research office tracing the published output of a specific award.
CF= (Conference) — conference titles, locations, dates, and sponsors.
FPY= (Final Publication Year) — distinguishes final publication from Early Access. A record with Early Access in 2018 and final publication in 2019 is returned by FPY=2019, and by either PY=2018 or PY=2019 — which is why a year-bounded systematic search can quietly double-count or miss Early Access items if you never think about the distinction.

One constraint that catches people out: PY= cannot stand alone. Clarivate documents that the year tag “must be combined with another field tag using the AND or NOT Boolean operator” — a bare PY=(2020-2024) is not a runnable query.

Address and Affiliation Tags

Institutional searches — building a publication list for a department, a REF/RAE return, or an annual research report — are their own problem, because institution names appear in records in dozens of variant forms. Web of Science exposes a graded set of tags for this:

OG= (Organization-Enhanced) — the one to reach for first. It searches Clarivate’s Preferred Organization Index, so a search on the preferred name returns records carrying that name and all of its known variants. This is what makes an institutional search tractable at all.
OO= (Organization) — literal, not expanded. A search on a preferred name returns records with that name; a search on a variant returns only that variant. Use it when you specifically want one spelling, not the variant cluster.
SG= (Suborganization) — departments, schools, institutes, and centres within a parent organisation.
CI= (City), PS= (Province/State), CU= (Country/Region), ZP= (Zip/Postal Code), and SA= (Street Address) — the geographic components of the address field.

A trap specific to organisation names: if the name contains a word Web of Science reads as an operator (AND, NOT, NEAR, SAME), it must be quoted or the query breaks. Clarivate’s own examples are OG=("Near" East Univ) and OG=("OR" Hlth Sci Univ).

Multiple field-tagged clauses combine with Boolean operators into one query, e.g.:

TS=("research data management" OR "data management plan") AND PY=(2022-2026) AND WC=("Information Science & Library Science")

Boolean and Proximity Operators

Web of Science recognises three Boolean operators. Contrary to a widespread claim in third-party search guides, they do not have to be typed in capitals — Clarivate’s current documentation states plainly that “search operators are not case sensitive; OR, Or, and or return the same results.” Capitals remain a sensible readability convention in a long query you intend to save, document, or hand to a reviewer, but they are a convention, not a requirement:

AND — both terms must appear (narrows results). Example: TS=(ORCID AND "author disambiguation").
OR — either term may appear (broadens results); use to combine synonyms and spelling variants. Example: TS=("data repository" OR "data archive" OR "data warehouse").
NOT — excludes records containing the second term. NOT is only available in Advanced Search, not Basic Search, and should be used cautiously — it can exclude relevant records that happen to mention the excluded term in passing.

Operator Precedence: the Order Web of Science Actually Evaluates Your Query

This is the single most common cause of a query that runs cleanly and silently returns the wrong set. When a query mixes operators, Web of Science does not evaluate left to right — it applies a fixed precedence order, highest first:

NEAR/x
SAME
NOT
AND
OR

Because AND binds more tightly than OR, an unparenthesised synonym list does not mean what it looks like it means. Clarivate’s own worked examples make the failure concrete: influenza OR flu AND avian finds records containing the word influenza, plus records containing both flu and avian — it does not require “avian” alongside “influenza.” Likewise copper OR lead AND algae returns everything mentioning copper, regardless of algae.

The remedy is mechanical, and worth applying unconditionally rather than case by case: parenthesise every OR list before combining it with anything else. Write TS=((influenza OR flu) AND avian). In a search you intend to save, publish in a PRISMA flow diagram, or defend to a reviewer, redundant parentheses cost nothing and remove the ambiguity entirely.

Two further defaults worth knowing, both documented: adjacent words with no operator are joined by an implicit AND (rainbow trout fish farm is equivalent to rainbow AND trout AND fish AND farm), whereas adjacent DOIs, PMIDs, or accession numbers are joined by an implicit OR, which is what makes pasting a list of identifiers into one search box work as a batch lookup.

Two proximity operators go beyond simple Boolean matching:

NEAR/x — finds terms within x words of each other in either order. Used without a number, NEAR defaults to a 15-word span. Example: TS=(researcher NEAR/5 identifier) finds records where “researcher” and “identifier” occur within 5 words of each other — tighter than AND (which ignores word distance entirely) and more flexible than a fixed phrase. NEAR/0 is a special case: it requires the terms to be adjacent. Two documented restrictions matter in practice — you cannot use AND inside a NEAR query (TS=(Germany NEAR/10 (monetary AND union)) is invalid, while nesting a second NEAR, TS=(Germany NEAR/10 (monetary NEAR/0 union)), is valid), and NEAR is not supported in the All Fields (ALL=) search. A quoted phrase or a parenthesised OR group is a valid NEAR operand: (salmon OR pike) NEAR/10 virus. If you actually want the literal word “near” in a title, quote it, or the query fails with an invalid-operator error.
SAME — restricts matches to terms appearing within the same sub-unit of a structured field. Clarivate defines its behaviour only for Address searches, where it requires the terms to occur within the same single address block rather than merely somewhere in the record: AD=(McGill Univ SAME Quebec SAME Canada). This is the difference between “this record has an author at McGill, in Quebec” and “this record mentions McGill somewhere and Quebec somewhere.” SAME is listed as available in Topic searches too, but Clarivate does not document what it means there, so treat Address as its reliable use. It is not supported in All Fields.

Wildcards and truncation work inside a field-tagged term, the same as most bibliographic databases:

* (asterisk) — replaces zero or more characters, e.g. biolog* matches biology, biological, biologist.
? (question mark) — replaces exactly one character, e.g. disrupt?r matches disrupter or disruptor.
$ (dollar sign) — replaces zero or one character, e.g. odo$r matches odor or odour.

Wildcards are more constrained than most database guides admit, and every one of these restrictions produces a silently wrong or failed search rather than a helpful error:

At least three characters must precede a wildcard in Topic and Title searches. zeo* is accepted; ze* is not.
Left-hand truncation is allowed in Topic, Title, and identifying-code searches (*bio), again with a three-character minimum after the wildcard for Topic and Title. It is not supported in Author or Cited Author searches, and not in All Fields.
Wildcards do not work in a publication-year search. PY=2007 and PY=(2012-2014) are fine; 200* errors.
Wildcards cannot follow special characters or punctuation (/ @ # . , : ; !).
Wildcards work inside a word, and you can mix them: l?chee$ matches lichee, lichees, lychee, and lychees.
Note that ? means exactly one character, not zero-or-one — Barthold? finds Bartholdi and Bartholdy but not Barthold. That is what $ is for.

Quotation marks force exact-phrase matching ("data management plan" as one phrase rather than three independently-matched words) — use this whenever a multi-word technical term should not be split apart by the search engine. Quotation marks do more than that, though, and the rest of what they do is the subject of the next section.

Lemmatization: Why One Word Returns Three

Web of Science does not search for the literal string you typed. By default it applies lemmatization — reducing a search term to its lexical root and retrieving the inflected forms alongside it. Clarivate’s documentation is explicit that this “can reduce or eliminate the need to use wildcards to retrieve plurals and variant spellings of a word.”

Concretely, and using Clarivate’s own examples:

cite also finds citing, cites, cited, and citation.
defense also finds defence — US and UK spelling variants are treated as the same lemma.

A closely related but distinct process, stemming, simply drops suffixes such as -ing and -es: a Title search for vinyl recording will find articles containing vinyl record. Clarivate describes the difference as stemming merely dropping suffixes, while lemmatization uses dictionaries defining clusters of words with shared meaning or morphology.

Three scope limits apply, all documented: lemmatization runs on Topic and Title searches; it works only with languages using the Latin alphabet; and both lemmatization and stemming apply only to English-language search terms.

How to Turn Lemmatization Off — and Why You Would

For most exploratory searching, lemmatization is helping you. For a systematic review, a bibliometric count, or any search whose result set has to be exactly reproducible and precisely bounded, silent term expansion is a problem: it inflates recall in a way that is invisible in the search string you report. There are two documented ways to switch it off.

Quotation marks. "mouse" finds records containing mouse but not mice; "color" finds color but not colour. Clarivate states that quotation marks turn off both lemmatization and the platform’s internal synonym finder.
The Exact Search toggle in the Advanced Search query builder. This is the one most third-party guides omit. Clarivate: “Exact Search is turned off by default so any search you perform will use stemming and lemmatization to expand your search… When Exact search is turned on, your search is limited to the exact term you type.” To run an exact search on a multi-word phrase with the toggle on, still put the phrase in quotation marks.

Quotation marks have one further documented side effect worth knowing in author searches: Web of Science normally appends an implicit trailing wildcard to a name, so AU=(Johnson M) behaves as Johnson M* and sweeps in Johnson MA, Johnson MB, and so on. Quoting it — AU=("Johnson M") — restricts the search to exactly Johnson M.

The Wildcard Interaction That Breaks Searches Silently

This is the most consequential interaction on the page, and it runs against intuition: using a wildcard turns lemmatization off for that term. Clarivate’s example is unambiguous — color* finds color, colors, and colorful, but not colour, colours, or colourful.

So the wildcard a careful searcher adds in order to be more thorough can quietly make the search less thorough, by suppressing the spelling-variant expansion that would otherwise have run. The documented fix is to write both variants explicitly and OR them: color* OR colour*.

The same applies to left-hand truncation: *valves returns bivalves but not bivalve; adding right-hand truncation as well, *valve*, returns both.

Lemmatization is also switched off automatically when an English-language query exceeds the number of terms allowed in a search — Clarivate documents the behaviour but does not publish the threshold, which is a reason to build very long synonym strings as explicit OR lists rather than relying on implicit expansion.

Hyphens, Punctuation, and Chemical Formulas

“Web of Science ignores hyphens” is the usual shorthand and it is wrong in a way that loses records. Clarivate documents a three-way split:

TS=hydro-power returns records containing hydro-power and hydro power — but not hydropower.
TS=hydro*power returns hydropower and hydroelectricpower.
TS=hydro power (no hyphen) applies the implicit AND and returns records with both words anywhere — hydro-electrical power, hydro-mechanical power, and so on.

Clarivate’s own advice is to enter a hyphenated term both with and without wildcards. For a systematic search, that means treating every hyphenated concept as at least two search lines.

Names behave differently again: hyphens and apostrophes in names are treated as spaces, so AU=(O Brien) and AU=(O'Brien) return the same count — though OBrien, closed up, may not. Ampersands are ignored in journal titles, so SO=(Past Present) matches Past & Present. Chemical formulas must not be spaced out: KxFe2-ySe2 works, while K x Fe 2-y Se 2 is parsed as a string of ANDed fragments and the quoted form returns nothing at all.

A Reproducibility Note for Systematic Reviewers

Clarivate’s Search Rules documentation carries a standing defect notice: between 20 July 2023 and 23 August 2023, quotation marks failed to disable lemmatization, so a query such as TS=("cat") wrongly matched both cat and cats. Clarivate advises anyone who ran a systematic search using quotation marks in that window to re-run it. If you are updating a review whose original search falls in that period, the search is worth re-running rather than carried forward.

Cited Reference Search: Tracing a Paper’s Citation History

Cited Reference Search is a distinct search mode from Basic and Advanced Search — it searches the reference lists inside indexed articles rather than searching articles’ own titles, abstracts, or keywords. It answers a different question: not “which papers are about this topic,” but “which papers have cited this specific work.”

The practical workflow:

Open Cited Reference Search and enter the cited author’s name and, ideally, the cited work’s title or an approximate publication year — Cited Reference Search fields typically include cited author, cited work/title, cited year, and cited source, since older or less-standardized reference lists don’t always cite consistently.
Web of Science returns a list of distinct “cited reference variants” it has matched — the same original paper can appear as several slightly different reference-list entries (a truncated title, a journal-abbreviation variant, a page-number discrepancy) because Web of Science indexes references exactly as they were printed in the citing article’s bibliography, not normalized against a master record.
Select the correct variant(s) — checking multiple variants that plausibly refer to the same source is often necessary to avoid undercounting — and run the search to retrieve every indexed article that cites that work.
The resulting citing-article list is itself a normal result set: it can be further filtered, exported, or combined with additional search criteria.

This is the mechanism behind forward citation tracking — starting from a known paper and finding everything that has cited it since, which is the core operation behind a citation-based literature review, a case for a paper’s ongoing influence in a tenure dossier, or confirming whether a retracted or corrected paper is still being cited without acknowledgment of the retraction. It is functionally the inverse of a reference list: a reference list shows what a paper cited going backward in time; Cited Reference Search shows what has cited a paper going forward in time.

Author Disambiguation: ResearcherID and ORCID

A plain AU= name search is inherently ambiguous: common names collide across many different real people, and the same person’s name may appear in a citing record under several variant forms (initials only, a former surname, a transliterated spelling). Two identifier layers exist specifically to make an author search precise instead of approximate:

Web of Science ResearcherID is Clarivate’s own author-profile identifier, introduced in 2008 (originally under Thomson Reuters), that lets a researcher claim their publication record within the Web of Science ecosystem. It is Clarivate-specific — useful for disambiguating and managing a profile inside Web of Science itself, but not portable to other publishers or systems the way an open identifier is.
ORCID iD is the non-proprietary, publisher-and-funder-agnostic persistent identifier for individual researchers (see CASRAI’s ORCID iD entry and its companion guide, How to Register, Find, and Search for ORCID iDs). Web of Science profiles can be linked to a researcher’s ORCID record, and Clarivate’s data ecosystem exchanges metadata with the ORCID registry so that publications claimed or verified on one side can inform the other.

Practically, this means: rather than (or in addition to) a name-based AU=(Smith J) search — which, for a common surname, can silently mix together dozens of unrelated researchers, or just as easily miss records where the name was entered as a variant — search or filter by the AI= (Author Identifiers) field using a known ResearcherID or ORCID iD, where one has been claimed and linked. Web of Science also applies its own machine-learning-assisted name-disambiguation to group probable-same-author records automatically where no external identifier is present, combined with human curation — but this is a best-effort algorithmic grouping, not a guarantee, and should not be treated as equivalent to a verified identifier match. For any use case where precision matters — a tenure dossier, a funder-mandated publication list, a systematic bibliometric analysis — search by identifier wherever one exists, and treat name-only author search results as a starting point to verify, not a final list.

Frequently Asked Questions

What is the difference between TS= and TI= in Web of Science?

TS= (Topic) searches the title, abstract, author keywords, and Keywords Plus together in one combined field. TI= (Title) searches only the article title. TS= casts a wider net and is the usual starting point; TI= is useful when a topic search returns too much noise and you want to require the term to appear specifically in the title.

How do I search Web of Science by a specific journal?

Use the SO= (Source/Publication Name) field tag in Advanced Search, e.g. SO=(Nature), optionally combined with other field tags using AND.

Can I exclude a term from a Web of Science search?

Yes, using the NOT Boolean operator in Advanced Search — NOT is not available in Basic Search. Use it cautiously, since excluding a term can also exclude relevant records that mention it only in passing.

What does NEAR/x mean in Web of Science?

NEAR/x is a proximity operator that finds two terms within x words of each other, in either order, inside the same field. Used without a number, NEAR defaults to a 15-word span. It sits between AND (which ignores word distance) and an exact phrase search (which requires an exact fixed order).

Do Boolean operators have to be capitalised in Web of Science?

No. Clarivate’s documentation states that search operators are not case sensitive — OR, Or, and or return the same results. Many third-party guides claim capitals are mandatory; they are a readability convention, not a requirement.

What order does Web of Science evaluate search operators in?

Highest precedence first: NEAR/x, then SAME, then NOT, then AND, then OR. Because AND binds more tightly than OR, an unparenthesised synonym list will not do what it looks like it does — always wrap an OR list in parentheses before combining it with AND.

How do I stop Web of Science expanding my search terms?

Put the term in quotation marks, or switch on Exact Search in the Advanced Search query builder. By default Web of Science applies lemmatization to Topic and Title searches, so cite also retrieves citing, cites, cited, and citation, and defense also retrieves defence. Quotation marks turn off both lemmatization and the internal synonym finder.

Why does adding a wildcard return fewer spelling variants?

Because a wildcard switches lemmatization off for that term. color* finds color, colors, and colorful, but not colour or colours — the spelling-variant expansion that lemmatization would have provided is suppressed. Write both forms and combine them: color* OR colour*.

Which field tag should I use for an institutional search?

OG= (Organization-Enhanced), not AD=. OG= searches Clarivate’s Preferred Organization Index, so it retrieves an institution’s name together with all its known variants. AD= searches the raw address field, which Clarivate notes is “usually for the first named author of an article” — so it will miss papers where your institution’s author was not listed first.

How is Cited Reference Search different from a normal topic search?

A topic search (TS=) searches the content of indexed articles themselves — their titles, abstracts, and keywords. Cited Reference Search searches the reference lists inside those indexed articles, letting you find every article that has cited a specific earlier work — forward citation tracking, rather than a keyword match.

Should I search by author name or by ResearcherID/ORCID?

By identifier whenever one is available and linked. A name-only AU= search is vulnerable to common-name collisions and name-variant misses; searching or filtering by AI= (Author Identifiers) using a ResearcherID or ORCID iD returns a precise match to that specific person’s claimed or verified record instead of an algorithmic best guess.

Related CASRAI Resources

Web of Science — what the platform is, its indexes, and its ownership history
Science Citation Index (SCI/SCIE) — the citation index at the core of the Web of Science Core Collection
Scopus vs Web of Science vs OpenAlex — a structured comparison of the three major discovery platforms
ORCID iD and How to Register, Find, and Search for ORCID iDs — the open identifier layer Web of Science profiles can link to
How to Look Up a Journal’s Impact Factor — the bibliometric indicator most closely associated with Web of Science’s Journal Citation Reports
Persistent Identifiers & Research Information Systems — the CASRAI cluster hub for this topic

Sources

Every field tag, operator, and default behaviour on this page is taken from Clarivate’s current first-party documentation rather than from third-party search guides — several of which state an uppercase-Boolean rule that Clarivate’s own text contradicts. Clarivate’s older webofscience.help.clarivate.com deep links now redirect to the help centre below.

Clarivate, Web of Science Core Collection Advanced Search Field Tags — the authoritative tag list, including AD=, OG=, OO=, SG=, CI=, PS=, CU=, ZP=, SA=, ALL=, UT=, PMID=, IS=, GP=, ED=, FO=/FG=/FT=, CF=, and FPY=.
Clarivate, Search Operators — operator precedence (NEAR/x, SAME, NOT, AND, OR), operator case-insensitivity, implicit AND and implicit OR, the NEAR/x default span of 15 words and NEAR/0, and the AND-inside-NEAR restriction.
Clarivate, Search Rules — wildcard semantics and truncation limits, lemmatization and stemming, the wildcard/lemmatization interaction, hyphen and punctuation handling, and the July–August 2023 quotation-mark defect notice.
Clarivate, Advanced Search Query Builder — the Exact Search toggle and its default state.
Clarivate, Web of Science Core Collection Search Fields — field-by-field operator availability, including the All Fields restrictions.
Clarivate, “Author data, made better together” (Web of Science ResearcherID / ORCID integration).

How to Search Web of Science Effectively