Form-based Event Capture


  • Data in event log is captured from an electronic form
    • User clicks Save to record entered data
  • All data items captured on the form have the same timestamp
    • Information about event ordering is ‘flattened’ into a single timestamp (time when the user clicked Save button)
  • Updating the form may result in all data items (not only changed data items) being re-written to the log
    • Apparent duplication of events


  • Complex discovered process models
    • Events with same timestamp treated as parallel events
    • Increased number of (and cross-cutting of) arcs on graphical models
  • Extraction of misleading process mining results
    • Due to inclusion of events that did not actually occur

Data Quality Issues

I16 - Incorrect data: timestamp, I27 - Irrelevant data: event
  • The temporal flattening introduced through the occurrence of this pattern negatively impacts the attribute accuracy of the log
    • Timestamps of the events reflect the saving of the form rather than the performance of the event
  • If the system records all fields on the form rather than only those fields that have changed, the trace completeness may be affected
    • Through erroneous inclusion of events that did not actually happen in the case

Manifestation and Detection

  • Pattern signature - groups of events, in the log with the same case identifier and same timestamp value
  • Marker events - existence of events with activity names similar to field labels known to exist on the same form


  • Simple — Aggregate all events, within each group of events having the same timestamp, into one event
    • Aggregated event represents the actual process step that took place
  • Alternate — Where events with similar timestamps contain two or more distinct and/or important process steps
    • Aggregate into two or more events reflecting the distinct process steps
    • NB: These new, aggregated events will share the same timestamp
  • Complex — An update to only some of the fields the form, triggers the recording of a new set of events, each with the same timestamp (some of which may be duplicate events because their values did not require any updates).
    • The updates of one or more data items in a form result in the recording of all fields as events in the log, or
    • The updates of one or more fields in a form will result in the recording of only those fields whose values have changed
      • Determine which fields have actually changed
      • Aggregate into one or more events reflecting the distinct process steps

Side-effects of Remedy

Where a process requires a form update, it may be the case that, for a given activity, the `new' value of a data item is the same as the `old' value of the data item. Where the form logging mechanism writes out all data values in the form, the fact that the `old' and `new' values of the data item are the same makes it dicult to determine whether the activity was carried out a second time or the data item was simply re-written as part of the form update process. In this scenario, we may lose the `update' action on those fields where `new' values and `old' values are the same.