Inductive databases (IDBs) contain both data and patterns. Here we consider IDBs where patterns are polynomial equations. We present a constraint-based approach to answering inductive queries in this domain. The approach is based on heuristic search through the space of polynomial equations and can use subsumption and evaluation constraints on polynomial equations. We evaluate this approach on standard regression problems. We finally consider IDBs containing patterns in the form of polynomial equations as well as molecular fragments, where the two are combined in order to derive QSAR (Quantitative Structure-Activity Relationships) models.
展开▼
