AI Glossary / Data science and analytics / Advanced

Boolean

Definition

Boolean refers to a logical value or expression with the two truth values true and false. Boolean logic combines conditions using operations such as AND, OR and NOT. In software, a Boolean type stores these logical values, although the handling of missing values and conversion from other types depends on the language or system.

What it means

A Boolean answers a precisely defined yes-or-no question. For example, an article may have a condition named references_checked, which is true after a specified review and false otherwise. The value does not explain the review process or establish that the article is accurate in every respect.
Boolean expressions combine such conditions to make decisions. A publishing rule might require both editorial approval and a completed reference check. The conditions must match the intended requirements. Correct logic can still produce an inappropriate decision if the inputs are stale, incorrectly recorded or too narrow for the purpose.

How it works

In ordinary two-valued logic, A AND B is true only when both inputs are true. A OR B is true when at least one is true, including when both are true. NOT reverses a truth value. Parentheses make the intended grouping of more complex expressions explicit.
Programming languages add implementation rules. Python supports truth-value testing of non-Boolean objects; for example, an empty string is false in a condition while a nonempty string is true. Its and and or operators can return operands rather than a Boolean value.
SQL systems also need to handle missing information. PostgreSQL permits null for an unknown Boolean state. Do not assume that unknown behaves like false, or that negating an unknown condition produces true. Check the semantics at each system boundary.

Three practical examples

• A content workflow enables publication only when both content approval and reference approval are true. It evaluates the approvals for the current revision rather than reusing an earlier version's status.
• A search interface combines category and level filters. Parentheses distinguish articles that are either introductory or intermediate within one category from a rule that accidentally includes intermediate articles from every category.
• A monitoring system records whether a service check passed. It treats an absent check result as unknown and investigates it separately from a confirmed failure, even if both require attention.

Think of it this way

Boolean logic resembles a set of clearly labeled gates. An AND gate opens when every required condition is satisfied; an OR gate opens when at least one permitted route is satisfied. The gate cannot inspect whether the labels are meaningful or the recorded conditions are trustworthy. Rule designers retain those responsibilities.

Worked example

Suppose a fictional article workflow uses two conditions: E means editorial approval and R means reference approval. Publication eligibility is E AND R. For the four possible input pairs, (false, false), (false, true) and (true, false) all produce false; only (true, true) produces true.
If someone accidentally uses E OR R, three of the four possible combinations produce true. That does not mean 75% of real articles will qualify: the four combinations need not occur equally often. It means the rule now accepts either approval alone as well as both approvals together.
Consider a separate Python import mistake. The text string "False" is nonempty, so testing its truth value gives true. That differs from the actual Boolean value False. An import routine should parse recognized text values explicitly and handle unrecognized or missing values according to a documented rule.

Illustrative case study

In a fictional project, Cedar Publishing imports approval records from a spreadsheet. The export stores the words “True” and “False” as text. An initial conversion checks whether each cell contains any text, which makes both words pass the condition.
The team corrects the conversion to recognize the allowed values and routes blanks or unexpected entries to an exception list. It then checks the publication rule against all four combinations of the two required approvals. The review also identifies approvals attached to old article revisions, which require separate handling even after the conversion is fixed.
Cedar now has a clearer connection between source values, logical conditions and publishing decisions. The example shows that type conversion and record context matter alongside a correct truth table. The platform and incident are fictional.

Common misconceptions and mistakes

OR ordinarily includes the case where both inputs are true. If exactly one must be true, that is an exclusive-or condition and should be stated explicitly. AND and OR are therefore not interchangeable shortcuts for combining requirements.
A false value is not automatically equivalent to missing information. Similarly, the strings “true” and “false” are text until interpreted under a conversion rule. Language-specific behavior can make casual conversions surprising.
A Boolean result is not a confidence score. A rule can return true even when its underlying evidence is weak, and converting a probability into true or false requires a threshold or decision rule that should be justified separately.

Business use

Booleans support workflow rules, validation, filtering and system configuration. Use names that express the condition clearly and document when a value is calculated or updated. Avoid ambiguous fields such as status_ok when different teams attach different meanings to them.
Review combinations at the boundaries of a rule, including missing values and stale records. For consequential actions, retain the evidence and reason behind the result rather than logging only true or false. When exchanging data between systems, check accepted encodings and null handling so the same record does not acquire a different meaning after transfer.

People also ask

Are Boolean values always stored as one bit?

No. Logical meaning and implementation storage size differ.

Is 1 always true and 0 always false?

They are common encodings, but conversion and type rules vary by system.

Can a Boolean prove a statement about the world?

It records or computes a truth value under a definition; the reliability of the underlying evidence remains a separate question.

Check your understanding

When is A AND B true?

Answer: When both conditions are true.

Does ordinary OR exclude both being true?

Answer: No.

Why is the Python string "False" risky in a simple condition?

Answer: It is nonempty text and therefore truthy, rather than the Boolean value False.

Related terms

• Binary variable
• Predicate (logic)
• Conjunction
• Disjunction
• Negation
• predicate
• truth table
• logical operator
• null
• type conversion
• conditional expression

Learn next

Write a truth table for a small approval rule, then add explicit cases for unknown inputs. Check how the chosen programming language or database represents each case before applying the rule to real records.

More in Data science and analytics

• Bar chart - A bar chart displays values using rectangular bars whose lengths represent magnitude on a shared numerical scale.
• Base effect - A base effect is the influence of the comparison period's value on a reported rate of change.
• Bernoulli trial - A Bernoulli trial is a random experiment whose recorded outcome has two possibilities, conventionally called success and failure.
• Box plot - A box plot summarizes the distribution of numerical data using a median, quartiles and whiskers.
• Categorical variable - A categorical variable records which category an observation belongs to, rather than a measured quantity.
• Causal analysis - Causal analysis investigates whether and how changing one factor would change an outcome.

Sources

Python documentation, Built-in Types, Truth Value Testing and Boolean Operations: language-specific conversion and operator behaviour. https://docs.python.org/3/library/stdtypes.html PostgreSQL documentation, Boolean Type: true, false and the null representation of unknown. https://www.postgresql.org/docs/current/datatype-boolean.html The approval rules and Cedar Publishing case are original illustrations. Implementation-specific behaviour is not a universal rule for all programming languages.

Previous: Bonferroni test Next: Boolean model of information retrieval