As part of a formal logic used to represent knowledge, the closed-world assumption (CWA) is the belief that an accurate statement is also known to be true. As a result, what is not now known to be true is untrue. 

The same word also refers to Raymond Reiter's logical formalisation of this assumption. The open-world assumption (OWA) is the inverse of the closed-world assumption, asserting that a lack of knowledge does not entail untruth. CWA vs OWA decisions impact understanding the true semantics of a conceptual expression using the same notations of ideas. A practical formalisation of natural language semantics can only sometimes avoid revealing whether the implicit logical backgrounds are CWA or OWA.

Negation as failure is connected to the closed-world assumption in that it entails believing untrue every predicate that cannot be proven true. The closed-world assumption is employed when it is certain that the knowledge base is exhaustive. Likewise, it is used when the knowledge base needs to be completed.

CWA applicability

When a system has comprehensive information, the CWA applies. It is true for a large number of database applications. Take, for example, a database application for airline reservations. If a direct flight between Austin and Madrid is not found in the database, the result is "There is no direct flight between Austin and Madrid." It is the expected and correct answer for this type of application.

OWA, on the other hand, is used when a system's information is incomplete. It is when we want to express knowledge and learn new things. For example, consider the clinical history system of a patient. If the patient's clinical history does not indicate a specific allergy, it is wrong to conclude that the patient does not have that allergy. Unless more information is provided to dispute the presumption, it is unknown if the patient has that allergy.

Data sources - researchers' perspective

A database's Closed-World Assumption (CWA) expresses that an atom not in the database is false. The CWA is only valid in domains where the database is entirely knowledgeable. In many circumstances, such as in distributed databases, a data source only has comprehensive knowledge of a subset of the discourse domain. 

The Belgium researchers provide an expressive and intuitively appealing approach to encoding a local closed-world assumption (LCWA) of autonomous data sources in their study. Their method distinguishes between data given by a data source and meta-knowledge about the domain in which the data is complete. In addition, the data is kept in a relational database. The meta-knowledge regarding its completeness is stated by a first-order theory, which can be processed by an independent reasoning engine (for example, a mediator).

Furthermore, the researchers consider several ways of representing their approach, compare it to existing techniques of capturing local closed-world assumptions about data sources, and demonstrate certain valuable aspects of their framework that ease its usage in real-world systems.