A class of Probabilistic Abductive Logic Programs (PALPs) is introduced and an implementation is developed in CHR for solving abductive problems, providing minimal explanations with their probabilities. Both all-explanations and most-probable-explanations versions are given. Compared with other probabilistic versions of abductive logic programming, the approach is characterized by higher generality and a flexible and adaptable architecture which incorporates integrity constraints and interaction with external constraint solvers. A PALP is transformed in a systematic way into a CHR program which serves as a query interpreter, and the resulting CHR code describes in a highly concise way, the strategies applied in the search for explanations.
Lecture Notes in Computer Science, 2008, Vol 5388, p. 85-118