Association acceleration for transaction databases

摘要

An association rule accelerator may be used to access a transaction database storing a plurality of transactions, each transaction including one or more items. The association rule accelerator also may select a sampling rate based on an item frequency of frequent items within the transaction database, relative to a sampled item frequency of sampled items within a corresponding sampled transaction database. An an association rule selector may determine, using the selected sampling rate and corresponding sampled transaction database, frequent item sets within the sampled transactions, and may further determine an association rule relating at least two items of the sampled transactions, based on the frequent item sets.

主权项

1. A system including instructions recorded on a non-transitory computer-readable medium, and executable by at least one processor, the system comprising:
an association rule accelerator configured to cause the at least one processor to access a transaction database storing a plurality of transactions, each transaction including one or more items, a view generator configured to cause the at least one processor to provide at least one visualization via a user interface in order to evaluate a plurality of sampling rates, each of the plurality of sampling rates corresponding to a different level of accuracy between the transaction database and a sampled transaction database, the at least one visualization depicting the levels of accuracy graphed relative to computation resources for the plurality of sampling rates; the association rule accelerator further configured to cause the at least one processor to receive a selection of a sampling rate among the plurality of sampling rates based on the at least one visualization, the association rule accelerator configured to cause the at least one processor to sample the transactions stored in the transaction database according to the selected sampling rate to obtain the sampled transaction database in a manner that corresponds to the selected sampling rate's level of accuracy between the transaction database and the sampled transaction database, wherein a number of sampled transactions in the sampled transaction database is less than a number of transactions in the transaction database; and an association rule selector configured to cause the processor to determine, using the selected sampling rate and the sampled transaction database, frequent item sets whose occurrence within the sampled transaction database is above a threshold level, and further configured to determine an association rule relating at least two items of the sampled transactions, based on the frequent item sets.