Structural domain modeling for understanding equipment failure messages

Demonstrates that domain-specific knowledge is required to recover implicit references to causality from narrative texts that describe equipment failures. Understanding texts that discuss complex pieces of equipment requires the possession not only of general knowledge about the types of objects and predicates in the domain, but also detailed knowledge about the particular equipment in question. This more expert level of knowledge is needed to dereference the names and descriptions of equipment referred to in the text and to infer their causal relationships and operational states when these are only implicitly expressed by the message writer. Knowing the structural configuration of the equipment is useful in both tasks, and a structural domain model can be extracted readily from equipment manuals and their accompanying parts lists, thus easing the knowledge acquisition bottleneck problem and making practical applications more feasible