A computer system takes as input an initial fault hierarchy KB0 and a set of annotated session transcripts C = {〈πj,rj〉} and is given a specified set of revision operators T = {θi} where each θi ∈ T maps a fault hierarchy KB to a slightly different hierarchy θi(KB). The computer system uses T to hill climb from the initial fault hierarchy KB0, through successive hierarchies, KB1... KBm, with successively higher empirical accuracies over C. At each stage, to go from a fault hierarchy KBk to its neighbor KBk+1, the computer system must evaluate KBk's accuracy over C, as well as the accuracy of each KB' ∈ N(KBk). The computer system provides an efficient way of evaluating the accuracy of KBk, and each θi(KBk), towards determining which, if any, θi(KBk) is more accurate than KBk. It exploits a few key observations. First, as each transformation used to map one hierarchy to a related one performs only local changes to the hierarchy, it will have only minor and easily computed effects on any instance. Second, most transformations will have no effect on many instances. Finally, one can bound how much the accuracy score for a knowledge base can change based on any instance, which means branch-and-bound techniques can be used to avoid computing the accuracy scores for various hierarchies that cannot be optimal.
展开▼
