Inductive inference with incompleteness

We present an axiomatic model of a process wherein likelihoods of eventualities are compared based on data. One eventuality is perceived as more likely than another whenever the data corroborates this conclusion. However, the correct relevance of records to the eventualities under consideration may be impossible to ascertain with any degree of surety due to multiple interpretations of the data, formalized by allowing the evaluator to entertain multiple weighting functions. The evaluator ranks one eventuality as more likely than another whenever its total weight over the entire database is higher, according to all relevance-weighting functions. Otherwise, the comparison is indecisive.