Tip #68 Searches Gone Wild! Popular Database Platforms and Their Wildcards
Written by Hilary Kraus and Zahra Premji
Take a walk on the wild side!
Join us for a deep dive into the world of wildcards. This post provides an introduction to how wildcards work, some potential pitfalls of using them, and detailed information about how to use them in many popular database platforms.
Wildcards 101
Complex searching often involves looking for various forms and spellings of a word. One common approach to this is the use of wildcards.
There are two varieties of wildcard. A mandatory wildcard replaces one letter; an optional wildcard replaces either zero or one letter. How these wildcards function and what syntax is used to operationalize them depends on where you're searching. In this post, we look at the ways commonly used databases (or database platforms) use wildcards.
Going forward, we will use root to refer to a collection of Latin alphabet letters to which a wildcard may be applied. Since wildcards are typically executed using a punctuation mark or special character, we'll use symbol to refer to wildcard syntax. Platforms differ in their rules about how many letters, symbols, or combinations of the two are permissible when using wildcards; we'll refer to those as characters. And we'll probably use database and platform interchangeably for brevity's sake (not that we are great at brevity, but it's good to have a goal).
Please note: truncation (which can be used to replace zero, one, or more than one letter) is another approach to searching for alternate word forms, but it's a whole other animal and will be addressed in a later post.
Hazards of going wild
Some caveats before we dive in:
Um, that's not the word I wanted.
Wildcards can add flexibility to your search, but they can also lead to unexpected noise. For instance, the following are examples of an optional wildcard (?) and a mandatory wildcard (#) on the Ovid platform. Note the benefits of using them, along with their potential pitfalls:
- colo?r??? finds color, colour, colorism, and also … Colorado
- wom### finds woman's, women's, and also … wombat
STOP! In the name of (stop)words …
Most databases or platforms have a standardized list of stopwords that are ignored when used in searches, either alone or as part of a phrase. These are typically terms like "by," "of," "the," or "to." Because databases treat stopwords differently from other words, the rules relating to wildcards discussed in this post may not apply. We recommend caution and careful testing before using wildcards to locate stopwords.
For example, in EBSCOhost, "of" and "been" are both stopwords. A search for "bee#" will find bees and beef, but not been. Searchers may often wish to find "of" within a phrase, such as "death of." No wildcard approach, however, (such as "death of#" or "death o?") will locate that phrase. Instead, "death" or "death of" finds the exact same number of results (simply, death), "death of#" finds death off, and "death o?" finds (incredibly) death-o-meter. For an example of how stopwords within phrases are treated on a different platform, and how a searcher might cope with the accompanying limitations, see the post on Stop Words within Phrases in Ovid.
A grave issue that can cause acute problems.
Depending on the database or platform, wildcards may apply only to Latin alphabet letters. We advise careful testing to be sure terms that have punctuation marks (women's) or diacritics (néonatalogie) are appearing in your search results when using wildcards to try to capture them. If you're not sure whether the wildcards are finding everything you expect, or testing indicates that you aren't, be sure you then include all the exact desired word forms, too.
Lega?, illega?? … it's a gr#y area.
In this post, we base our recommendations upon platform or database documentation combined with extensive testing (and sometimes on follow-up conversations with customer support). Unfortunately, there are lots of possibilities not addressed in help files, including things like stacking wildcards (that is, using multiple in a row without letters in between: colo?r???). We've done our best to confirm and clearly state what works and what doesn't, whether it matches the documentation, and what things we do not recommend regardless.
Literally no one asked for this.
Some platforms automatically look for alternate word forms (such as performing stemming or even lemmatization), often with little transparency. This may or may not be overridden by the use of wildcards. It is essential to check the documentation thoroughly when designing searches, even if you are not using wildcards, as these default behaviors (or behaviours lol) may impact your results. The details of stemming, lemmatization, and other fancy footwork these platforms or databases do without our intervention is beyond the scope of this post.
Ready to go wild?
Following are guidelines for how to execute wildcard searching in several commonly used databases and platforms, as well as links to relevant search documentation.
Bookmark the table:
You can also bookmark our handy Wildcard Syntax Table (Google Docs version), or the matching Wildcard Syntax Table (Google Sheets version) in our Search Translation Toolkit, that summarizes the essentials!
Cochrane Library (Wiley)
Documentation
Optional wildcard, replaces zero or one letter
Symbol: ? (question mark)
Usage:
- May be used at the beginning, internally, or at the end of a root: ?co or colo?r or color?
- Our testing of using multiple optional wildcards in the same root led to inconsistent results, and Cochrane's help desk says this is not formally supported, so we recommend careful testing before using wildcards here (or simply including all possible desired word forms and avoiding use of wildcards altogether).
Notes:
- Cochrane Library requires a minimum of three characters (not letters) in a root containing wildcards; wildcards are counted as acceptable characters: v?p
- Wildcards may not be used in phrases (within quotation marks); to use wildcards in a phrase, use the NEXT operator instead, in the format root NEXT root: left NEXT arm?
EBSCOhost
Documentation
Mandatory wildcard, replaces exactly one letter
Symbol: ? (question mark)
Usage:
- May be used internally in a root, but not at the beginning: wom?n
- To use a mandatory wildcard at the end of a root, precede it with a #: phobi#? Our testing showed that root#? returns the root plus either 1 or 2 letters (e.g., phobia, phobic, phobias), so it's not precisely equivalent to a standard mandatory wildcard.
- Multiple mandatory wildcards may be used in the same root and/or stacked together: wom???s
Optional wildcard, replaces zero or one letter
Symbol: # (hash/pound sign)
Usage:
- May be used internally or at the end of a root, but not at the beginning: colo#r or color#
- Multiple optional wildcards may be used in the same root and/or stacked together: colo#r# or colo#r###
Notes:
- EBSCOhost requires a minimum of one letter in a root before an optional wildcard: v#
- A mandatory wildcard requires a minimum of two letters, one on either side of the wildcard, because a mandatory wildcard may not be used alone at the end of a root: v?p
- Mandatory and optional wildcards may be used in the same root: vapo#ri?e
- To use multiple wildcards, the root must include a minimum of four characters, with a minimum of two of those characters being letters: v?p# or v#p# or v?#p (etc.) but not v??#
- Searches containing violations of these rules will run, rather than throwing an error message, but the results will ignore the wildcards, so v? or v??# will return results for, simply, v
Embase.com (Elsevier)
Documentation
Mandatory wildcard, replaces exactly one letter
Symbol: ? (question mark)
Usage:
- May be used at the beginning, internally, or at the end of a root: ?omen or wom?n or phobi?
- Multiple mandatory wildcards may be used in the same root and/or stacked together: wom???s
Optional wildcard, replaces zero or one letter
Symbol: $ (dollar sign)
Usage:
- May be used at the beginning, internally, or at the end of a root: $co or colo$r or color$
- Multiple optional wildcards may be used in the same root and/or stacked together: colo$r$$$
Notes:
- Embase.com requires a minimum of three letters in a root before a wildcard: vap$
- Mandatory and optional wildcards may be used in the same root: vapo$ri?e
- Searches using the optional wildcard are automatically translated by Embase.com as [root without wildcard] or [root with mandatory wildcard]: colo$r becomes color or colo?r
Europe PMC
Europe PMC's search syntax guide has a section labeled "Wild card & truncation," but their definition of wildcard does not match the one we use in this post.
Lens.org
Documentation
Mandatory wildcard, replaces exactly one letter
Symbol: ? (question mark)
Usage:
- May be used at the beginning, internally, or at the end of a root: ?omen or wom?n or phobi?
- Multiple mandatory wildcards may be used in the same root and/or stacked together: wom??? or ph?bi??
Notes:
- To use wildcards, it is critical to first deactivate stemming, which is on by default. Stemmed searching causes wildcards to behave in unexpected and inconsistent ways, and is thus incompatible with effective, replicable systematic searches. To turn off stemming, go to Structured Search → Query Tools → uncheck the Stemmed box.
- Ok, you turned off stemming, right? You may proceed.
- Lens requires a minimum of one letter in a root before or after a wildcard: ?v or v?
- Wildcards may not be used within quotation marks, whether enclosing a single word or a phrase.
OpenAlex
Documentation
Mandatory wildcard, replaces exactly one letter
Symbol: ? (question mark)
Usage:
- May be used internally, but not at the beginning or end of a root: wom?n
- The documentation does not clarify if multiple mandatory wildcards may be used in the same root and/or stacked together.
Notes:
- OpenAlex's documentation is extremely limited, and our testing sometimes showed different results than would be expected.
- While OpenAlex's documentation says a "search term must have at least 3 characters before the wildcard," our testing showed a minimum of one letter in a root before and after a mandatory wildcard will work: v?p
- Our testing of using multiple mandatory wildcards in the same root led to inconsistent results.
- Because of all this, we urge you to test your own searches thoroughly before using wildcards in OpenAlex, or simply include all desired word forms and avoid the use of wildcards altogether.
Ovid
Documentation
- Truncation and Wild Cards
- For you MEDLINE fans, here's the MEDLINE documentation link. (The wildcard information is no different from the default documentation, but you might prefer it in your bookmarks since it has all the field codes for MEDLINE)
Mandatory wildcard, replaces exactly one letter
Symbol: # (hash/pound sign)
Usage:
- May be used internally or at the end of a root, but not at the beginning: wom#n or phobi#
- Multiple mandatory wildcards may be used in the same root and/or stacked together: wom### or ph#bi##
Optional wildcard, replaces zero or one letter
Symbol: ? (question mark)
Usage:
- May be used internally or at the end of a root, but not at the beginning: colo?r or color?
- Multiple optional wildcards may be used in the same root and/or stacked together: colo?r? or colo?r???
Notes:
- Ovid requires a minimum of one letter in a root before a wildcard: v# or v?
- Mandatory and optional wildcards may be used in the same root: vapo?ri#e
ProQuest
Documentation
Optional wildcard, replaces zero or one letter
Symbol: ? (question mark)
Usage:
- May be used internally or at the end of a root, but not at the beginning: colo?r or color?
- Multiple optional wildcards may be used in the same root and/or stacked together: colo?r? or colo?r???
Notes:
- ProQuest requires a minimum of one letter in a root before a wildcard: v?
PubMed
PubMed offers a "wildcard," but it operates as truncation, replacing zero or more letters. Since it has no traditional mandatory or optional wildcard, it is outside the scope of this post.
Scopus
Documentation
Mandatory wildcard, replaces exactly one letter
Symbol: ? (question mark)
Usage:
- May be used at the beginning, internally, or at the end of a root: ?omen or wom?n or phobi?
- Multiple mandatory wildcards may be used in the same root and/or stacked together: wom??? or ph?bi??
Notes:
- Scopus requires a minimum of one letter in a root before a wildcard: v?
- Because Scopus treats hyphens as spaces (except when { } are used), a search for a root with a wildcard may find the root standing alone or as part of a hyphenate: ?p finds bp, but also miR-23a-3p, RP-HPLC, and follow-up
- Wildcards can be used in phrases when using quotation marks ("loose phrase" searching): "water vapo???"
Web of Science
Documentation
Mandatory wildcard, replaces exactly one letter
Symbol: ? (question mark)
Usage:
- May be used at the beginning, internally, or at the end of a root: ?omen or wom?n or phobi?
- Multiple mandatory wildcards may be used in the same root and/or stacked together: wom???
- In Title and Topic searches, there must be at least three consecutive letters within the root: AB=(ph?bi??) or TI=(ph?bia?) but not TI=(ph?bi??)
Optional wildcard, replaces zero or one letter
Symbol: $ (dollar sign)
Usage:
- May be used at the beginning, internally, or at the end of a root: $olor or colo$r or color$
- Only one optional wildcard may be used in a root
- In Title and Topic searches, a root must have a minimum of three letters, but those letters need not be consecutive: TS=(vap$) or TS=(va$p) or TS=($aps)
Notes:
- Web of Science usually requires a minimum of one letter in a root before a wildcard: v? or v$ ; exceptions when using mandatory or optional wildcards in Title and Topic searches are described above.
- Mandatory and optional wildcards may be used in the same root: vapo$ri?e
Comments
Post a Comment