Attributional Calculus A Representation System and Logic for Deriving Human Knowledge from Computer Data

Attributional calculus is a typed logic system that combines elements of propositional logic, predicate logic, and multiple valued logic. It was developed with the main intention to support natural induction, by which we mean an inductive inference that generates knowledge in the forms “natural” to people, such as natural language style descriptions or graphical visualizations, and by that simple to interpret and easy to relate to human knowledge. To this end, attributional calculus employs new descriptive constructs and operators that can significantly simplify descriptions that, if represented using standard logic operators, would be complex and opaque. Sentences in attributional calculus can be interpreted as binary logic expressions (crisp interpretation), or as multiple valued or continuously valued logic expressions (flexible interpretation). Attributional calculus stems from variable valued logic (VL1), and its various subsets have been be implemented in the AQ type learning programs. Conventional decision rules and association rules employed in data mining can be viewed as special cases of attributional rules.