| 摘要 |
A system and method to facilitate automatic identification of event classification errors in a network are described. Session data containing events logged by a user entity over a network in a predetermined period of time is retrieved from one or more event logs. Each event is further parsed to generate one or more event units. A frequency parameter of co-occurrence of the event units within each event of the session data is further determined and at least one session distance is determined among the events based on the frequency parameter. At least one classification distance is further retrieved from a data storage module, such as, for example, a database or a datastore, the one or more classification distances representing a relation between the events and corresponding classification categories within the database or datastore. Each session distance is further compared to each retrieved classification distance to ascertain event classification errors within the database or datastore. A predetermined error code is then assigned to each event if the corresponding session and classification distances are different. Finally, a list of classification errors containing the identified events and their corresponding error codes is output for further review and analysis, either through manual editing or, in the alternative, through automatic editing.
|
