How to Search Academic Databases Like a Researcher: Keywords, Boolean Operators, and Search Strategies

How do researchers turn a topic into an effective database search?

You know what you want to research. You open an academic database, type the topic into the search box, and either get thousands of irrelevant results or almost nothing useful.

The problem is often not the topic. It is the way the question has been translated into a database query.

Academic databases do not interpret a research problem exactly as a human researcher does. Effective searching requires you to identify the concepts in your question, anticipate the different language authors may use for those concepts, and tell the database how those terms should relate to one another.

That means learning a small but powerful vocabulary of search techniques: keywords, phrases, Boolean operators, parentheses, truncation, wildcards, field searching, controlled vocabulary, and, where supported, proximity operators. The syntax varies among databases, so a good search strategy is normally adapted rather than copied unchanged from one platform to another.

Build searches from concepts, not complete questions

The basic logic is simple. OR usually broadens a search by accepting alternative terms, while AND usually narrows it by requiring different concepts to occur. NOT excludes records and should be used cautiously because it can remove relevant material as well as irrelevant material.

How to build an academic database search from the ground up

Step 1: Stop searching your research question as a sentence

Suppose your research question is:

How does remote work affect job satisfaction among software developers?

Searching that entire sentence is usually not the best starting point. Instead, identify the concepts carrying the meaning of the question:

• remote work;
• job satisfaction; and
• software developers.

These become concept groups. You can then develop different ways authors might express each one.

This concept-based approach gives you control over the search. If one idea is poorly represented in the results, you can improve that concept group without rebuilding everything else.

Step 2: Think beyond your preferred keyword

A keyword is useful only if authors use it. Your preferred terminology may not be the terminology used by another discipline, an older paper, another country, or the database's indexing system.

For each important concept, consider:

• synonyms;
• abbreviations and acronyms;
• alternative spellings;
• older and newer terminology;
• broader or narrower terms when appropriate;
• discipline-specific terminology;
• related expressions that represent the same concept; and
• controlled-vocabulary terms used by the database.

Cochrane's search guidance, for example, recommends considering synonyms, related terms, variant spellings, text words, and controlled vocabulary when developing concept groups.

One of the best places to discover vocabulary is the literature itself. Find several highly relevant papers and inspect their titles, abstracts, author keywords, and database subject terms. The language used in good seed papers can expose terminology your initial brainstorming missed.

Step 3: Use OR to combine alternative terms

OR tells the database that any of the connected alternatives may satisfy the concept.

For example:

("remote work" OR telework OR telecommuting OR "working from home")

A record does not need to contain every term. It needs to match at least one of the alternatives.

This is why OR normally broadens retrieval. Adding another genuine synonym gives the database another way to recognize the concept. Cochrane recommends normally combining terms representing the same concept with OR.

Step 4: Use AND to connect different concepts

AND performs a different job. It requires the connected concepts to be represented.

For example:

("remote work" OR telework OR telecommuting) AND ("job satisfaction" OR "work satisfaction")

The first parentheses represent one concept and the second represent another. AND asks for records satisfying both groups.

Adding another AND condition generally makes retrieval more restrictive. That can improve relevance, but it can also remove useful papers if the extra concept is not consistently mentioned or indexed.

Step 5: Use parentheses to make your logic explicit

Once AND and OR appear in the same query, grouping matters.

Compare:

"remote work" OR telework AND "job satisfaction"

with:

("remote work" OR telework) AND "job satisfaction"

The second query explicitly tells the database that “remote work” and “telework” belong to one concept group and that the group must be combined with job satisfaction.

Do not assume every database evaluates mixed operators from left to right. Scopus, for example, applies a documented operator order of precedence in Advanced Search. Parentheses make your intended logic clearer and reduce dependence on remembering a platform's default processing rules.

Step 6: Treat NOT as an exclusion tool, not a cleanup button

NOT, or an equivalent exclusion operator such as Scopus's AND NOT, removes records containing specified terms.

Suppose a search for jaguar retrieves large numbers of papers about the animal when you need research on the automobile brand. Exclusion may appear attractive.

The danger is that a relevant paper can contain both the wanted concept and the word you exclude. A paper comparing vehicle-inspired biomimetic design with animal locomotion, for example, might disappear even if part of it is relevant.

Step 7: Use quotation marks when words need to stay together

Phrase searching can distinguish a concept from separate occurrences of its component words.

Searching:

"climate change"

is different from simply entering:

climate change

depending on the database and interface.

Quotation marks are therefore useful for multiword concepts such as “job satisfaction,” “machine learning,” or “social media.” But phrase behavior is database-specific. Scopus, for example, distinguishes a loose phrase in double quotation marks from an exact phrase enclosed in braces.

Do not assume quotation marks have precisely the same effect everywhere. Check the database's current help documentation when exact behavior matters.

Step 8: Use truncation to capture useful word endings

Truncation replaces part of a word with a symbol so that multiple word forms can be retrieved.

In a system where the asterisk is the truncation symbol:

educat*

might help retrieve forms beginning with the same stem, such as education, educational, and educator.

This can reduce the need to type every variation separately, but truncating too early in a word may retrieve large numbers of unrelated terms.

Cochrane notes that truncation and wildcard conventions vary across database interfaces, including where the symbol can appear and how many characters it can represent. A search using random* in one interface therefore should not automatically be assumed to behave identically elsewhere.

Step 9: Use wildcards for spelling variations when appropriate

A wildcard usually substitutes for one or more characters within a term. It can help when words have predictable spelling variations.

Cochrane gives the example:

wom?n

to represent both “woman” and “women” in interfaces where the question mark functions as the appropriate wildcard.

But wildcard symbols are not universal. Scopus currently uses * for zero or more characters and ? for a single character. It also performs forms of word stemming or lemmatization in many text searches, meaning some variations may already be retrieved without truncation.

Use wildcards because you know what variants you need, not simply because a search looks more sophisticated with them.

Step 10: Search specific fields when a general search is too noisy

Academic records contain separate bibliographic fields: title, abstract, keywords, author, affiliation, publication year, source title, DOI, and others.

Field searching lets you tell the database where a term must occur.

If a broad search retrieves many papers that mention your term only incidentally, restricting it to titles, abstracts, or keywords may improve relevance. If you need one known paper, searching its DOI may be more precise than searching its title words.

Scopus, for example, allows document searching within specified fields and Advanced Search supports field codes. Its commonly used TITLE-ABS-KEY field searches article titles, abstracts, and keywords.

The available fields and their codes differ by database. Translate the underlying idea, not necessarily the literal field code.

Step 11: Understand controlled vocabulary

Keywords are the words you supply. Some bibliographic databases also use controlled vocabulary: standardized subject terms used to represent concepts consistently.

These approaches complement each other. A new concept may appear in titles and abstracts before an indexing vocabulary fully captures it, while controlled vocabulary can connect papers whose authors describe the same idea using different language.

When the database provides controlled vocabulary and your project requires a thorough search, investigate both approaches rather than assuming keywords alone capture every relevant record.

Step 12: Use proximity searching when phrases are too rigid

Sometimes you want two concepts to occur near each other without requiring one exact phrase.

An author might write “training provided to nurses,” “training of hospital nurses,” or “nurse training.” A rigid phrase search could miss useful variants, while a simple AND search might retrieve records where the two words appear far apart and have little relationship.

Proximity operators provide a middle ground by specifying how close terms must appear. Their syntax varies considerably by platform. Scopus supports W/n, where order is unrestricted, and PRE/n, where one term must precede the other. For example, Scopus documents journal W/2 publishing as a search in which the terms occur within the specified distance.

Other interfaces use different operators. Cochrane's technical search guidance documents examples including ADJ in Ovid, N in EBSCOhost, and a separate proximity syntax in PubMed.

Never transfer a proximity expression to another database without checking its syntax and definition of distance.

Step 13: Build the search one concept at a time

Long search strings are easier to troubleshoot when constructed in stages.

A practical workflow is:

1. search the first concept and its synonyms;
2. inspect the result count and sample records;
3. build the second concept;
4. test it independently;
5. combine the two concept groups with AND;
6. inspect what changed;
7. add another concept only when it is genuinely necessary; and
8. continue testing after each major change.

This makes it much easier to identify why a search suddenly produces zero results or thousands of irrelevant ones. Scopus likewise recommends constructing large searches carefully and notes that complex combinations of common terms, OR groups, wildcards, filters, and proximity operators can become computationally expensive or ambiguous.

Step 14: Test your strategy against known relevant papers

If you already know several papers that clearly belong in your literature set, search for them directly and then check whether your general strategy retrieves them.

If an important paper is missing, investigate why:

• Does it use a synonym you omitted?
• Is your phrase search too restrictive?
• Did an AND condition eliminate it?
• Is an exclusion term removing it?
• Are you searching the wrong field?
• Does the database index the paper at all?

A missing known paper does not automatically prove that the entire strategy is bad, but it gives you a concrete diagnostic problem to investigate. You can also use citation chaining to discover relevant research that a keyword-based strategy may have missed.

Step 15: Translate the strategy when you change databases

A concept strategy can travel between databases. Its exact syntax often cannot.

You may keep the same basic concepts and synonyms while changing:

• field codes;
• truncation symbols;
• wildcards;
• phrase syntax;
• proximity operators;
• controlled-vocabulary terms;
• operator conventions; and
• available filters.

Cochrane specifically advises checking database help files when adapting searches because truncation and wildcard functionality differs between interfaces.

If you are still deciding which discovery and citation database belongs in your workflow, compare the coverage and functionality you need. For the broader workflow around selecting sources, documenting searches, and organizing results, consider the literature search as a whole.

Building a database search step by step

The final string is not automatically the “correct” search. It is a testable search strategy. Its quality depends on whether the concepts and terms represent the question appropriately and whether the database interprets the syntax as intended.

Search techniques that look sophisticated but can make results worse

Use search operators to solve specific retrieval problems

You do not need every advanced search feature in every query. Start with the simplest structure that represents your research question, then introduce additional techniques when the results show why you need them.

A well-designed search is not necessarily a complicated search. The objective is to express your research concepts clearly enough that the database retrieves useful evidence without imposing unnecessary restrictions.

Academic database search checklist

Frequently asked questions about academic database searching

Search strategy matters as much as the words you type

How to Cite This Guide

This guide is intended to be read, shared, and used in research, teaching, and academic work. If you draw on its ideas, explanations, or other content, please acknowledge the source by citing the guide. Doing so gives appropriate credit and helps your readers locate the original resource.