Pattern-based stability analysis of complex data sets

摘要

Methods and systems for identifying stability exceptions in a data log are disclosed. In one method, at least one key that is present in the data log is determined. The data log is comprised of at least one data set, at least one of which includes a plurality of iterations indicating states of the corresponding data set at different points in time. For each data set and for each key, a map is generated. The map indicates, for each iteration of the corresponding data set, whether the corresponding key is present in the corresponding iteration. Moreover, at least one expression pattern rule that models data item stability characteristics over data set iterations is compared to each of the maps to determine whether the corresponding map satisfies the one or more expression pattern rules. Further, at least one unstable data item is identified in the data log based on the comparison.

主权项

1. A method for identifying stability exceptions in a data log comprising:
determining at least one key that is present in the data log, wherein the data log is comprised of at least one data set, at least one of which includes a corresponding plurality of iterations indicating binary transition states of the corresponding data set at different points in time, and wherein each key of the at least one key denotes a different, respective data item in the data log; for each data set of the at least one data set and for each key of the at least one key, generating a map indicating, for each iteration of the corresponding data set, whether the corresponding key is present in the corresponding iteration; determining adherence, by a processor, of at least one expression pattern rule that models data item stability characteristics over data set iterations to the map generated for each data set and for each key to determine whether the corresponding map satisfies the at least one expression pattern rule; and identifying at least one key exception as at least one unstable data item, respectively, in the data log based on the determining adherence.