Using Domain Models for Data Characterization in PBE

INTRODUCTION One of the main problems in PBE inferencing is that of deriving abstract data characterizations from concrete objects and values involved in user’s manipulations [8]. This involves selecting variables and constants (parameterization) among involved values, and establishing what variables represent. Some variables will be input parameters of the construct or procedure inferred by the PBE system, and some will be expressions that indicate values or objects that are obtained from input parameters by traversing a data structure made of relations, lists and attributes, and operations on them. Our contribution to this workshop sets the focus on the use of rich descriptions of application data to achieve correct and expressive data characterizations.