predicatesgain

Predicatesgain is a concept in knowledge representation and logic that denotes the growth of the predicate vocabulary in a system as a result of inference, definitions, or data augmentation. It concerns how a base knowledge base can expand its repertoire of predicate symbols beyond the initial set.

Formalization can be described by considering an initial set of predicate symbols P0. After applying inference steps, the system yields sets P1, P2, and so on. The predicatesgain at step t can be defined as the difference |Pt| − |Pt−1|, capturing new predicate symbols that become available. Some authors distinguish between predicate-symbol gain (new predicate names) and predicate-instance gain (new facts that use existing predicates). Predicatesgain is thus a measure of vocabulary growth, not merely of new facts.

Examples illustrate the idea: a rule that derives is_ancestor(X,Z) from parent(X,Y) and parent(Y,Z) introduces a new predicate symbol, contributing to predicate-symbol gain. If a rule adds a higher-order predicate like related(P,Q) that aggregates multiple relations, that also counts as predicatesgain.

Applications and implications include studying expressivity, reasoning complexity, and potential predicate explosion in knowledge bases and rule-based systems. Understanding predicatesgain helps in designing mitigation strategies such as arity or type restrictions, stratification, or limiting rule schemas to control vocabulary growth.

History and usage: The term predicatesgain is not widely standardized and may appear in niche AI literature as a descriptive notion of vocabulary growth in reasoning systems. See also logic programming, knowledge representation, rule-based systems, predicate calculus, and predicate explosion.