| 摘要 |
In one aspect, the invention relates to a method for clustering a plurality of subsequent records (102) comprising a time field with a time stamp (103). In a first step the records (102) are grouped into a sequence of bursts (101). The difference between the time stamp of two subsequent records within a burst is no more than a predetermined time difference T1 and the difference between the time stamp of two subsequent records of subsequent bursts is equal or more than this predetermined time difference T1. In a second step, the bursts are grouped into a sequence of sessions (104). A difference between the time stamp of two subsequent records belonging to subsequent bursts of a same session is no more than a predetermined time difference T2. A difference between the time stamp of subsequent records belonging to subsequent bursts of subsequent sessions is equal or more than the predetermined time difference T2. |
