MOMO-model-based diagnosis for everybody

It is shown how to implement the core of a model-based diagnosis system by a small hyperresolution-based procedure using Prolog. The algorithm is able to find all possible diagnosis candidates. The model-based diagnosis by model generation (MOMO) algorithm has well-defined semantics and allows the description of models using general range-restricted clauses, in contrast to earlier systems, which only allow a Horn clause description. A large class of systems can be modeled and can incorporate different types of behavioral models, such as correct behavior models, alibis and physical necessity rules. As the basic algorithm can be easily implemented by a few Prolog clauses, it can serve as a testbed for various ideas concerning model-based diagnosis without using a full-fledged environment incorporating assumption-based truth maintenance system (ATMS) techniques. The algorithm has been tested using several models including well-known example systems as well as various modeling assumptions.<
>