Abstract
The O[M] formalism for Order-of-Magnitude reasoning is described. O[M] is based on seven primitive relations among absolute magnitudes of quantities: 'much less than' (< <), 'moderately less than' (- <), 'slightly less than' (∼ <), 'equal to' (= =), 'slightly greater than' (> ∼), 'moderately greater than' (> -), and 'much greater than' (> >). 21 compound relations are formed as implicit disjunctions of consecutive primitive relations. A strict interpretation of the relations allows exact conservative inferences, while a heuristic interpretation allows more aggressive and human-like inferences, by permitting some slack at each inference step. Inference strategies are based on propagation of order-of-magnitude relations through properties of the relations, algebraic constraints, and rules. O[M] operates mainly in the data-driven direction with assumption-based truth-maintenance for the resolution of contradictions. O[M] provides efficient integration of quantitative and qualitative knowledge in the expression and solution of engineering problems. The system has been applied in process engineering and biochemical engineering.
Original language | English (US) |
---|---|
Pages (from-to) | 106-114 |
Number of pages | 9 |
Journal | Artificial Intelligence in Engineering |
Volume | 4 |
Issue number | 3 |
DOIs | |
State | Published - Jul 1989 |
Externally published | Yes |
Keywords
- approximate reasoning
- approximate relation
- approximation
- artificial intelligence
- order-of-magnitude reasoning
- qualitative reasoning
- relative magnitude
- semiquantitative reasoning
ASJC Scopus subject areas
- Computer Science(all)
- Engineering(all)