"... In PAGE 5: ...work)case (i.e., a process instance), (iii) each event can have a performer also referred to as originator (the person executing or initiating the activity), and (iv) events have a timestamp and are totally ordered. Table2 shows an example of a log involving 11 events, 9 activities, and 4 performers. In addition to the information shown in this table, some event logs contain more information on the case itself, i.... In PAGE 5: ...Table 2: An event log. Event logs such as the one shown in Table2 are used as the starting point for mining. We distinguish three different perspectives: (1) the process perspective, (2) the organizational perspective, and (3) the case perspective.... In PAGE 6: ... To illustrate the first two consider Figure 1. The log shown in Table2 contains information about two cases (i.e.... In PAGE 6: ... Each case starts with the execution of A and B. Based on the information shown in Table2 and by making some assumptions about the completeness of the log (i.e.... In PAGE 6: ... Logging start- and end events would form a good base for this parallelism detection. The number of events in Table2 is too small to establish our assumptions accurately. However, for the sake of argument we assume that the things that did not happen will never happen.... In PAGE 6: ... However, for the sake of argument we assume that the things that did not happen will never happen. Note that there may be may models that generate the log shown in Table2 .... In PAGE 6: ...igure 1(a) does not show any information about the organization, i.e., it does not use any information on the people executing activities. However, Table2 shows information about the performers. For example, we can deduce that activity B is executed by either Peter or Monika, and that all other activities are executed by a single specified person.... In PAGE 7: ... Figure 1 (b) and (c)) or on the utilization of performers or execution frequencies. This example process (Figure 1, Table2 ) is further used in the remainder of this section, but then we will use real activity names instead of A, B etc. To address the three perspectives and the logical and performance issues we have developed a set of tools including EMiT [2,11], Thumb [29], MinSoN [4], and the ProM tool.... ..."

"... In PAGE 6: ...he process), (ii) each event refers to a case (i.e. a process instance), (iii) each event can have a performer also referred to as originator (the person executing or initiating the activity), and (iv) events have a timestamp and are totally ordered. Table1 shows an example of a log involving 19 events, 5 activities, and 6 originators.1 In addition to the information shown in this table, some event logs contain more information on the case itself, i.... In PAGE 6: ... For example, the case handling system FLOWer logs every modification of some data element. Event logs such as the one shown in Table1 are used as the starting point for mining. We distinguish three different perspectives: (1) the process perspec- tive, (2) the organizational perspective and (3) the case perspective.... In PAGE 6: ... For example, if a case represents a replenishment order it is interesting to know the supplier or the number of products ordered. 1 Note that in Table1 we abstract from event types, i.e.... In PAGE 8: ... To illustrate the first two consider Figure 2. The log shown in Table1 contains information about five cases (i.... In PAGE 8: ... However, for some cases activity C is executed before activity B. Based on the information shown in Table1 and by making some assumptions about the completeness of the log (i.e.... In PAGE 8: ...elations between the various properties of a case (i.e., the What? question). Focusing on the case perspective is most interesting when also data elements are logged but these are not listed in Table1 , and therefore this perspective is not addressed in Figure 2. Note that some of the properties may refer to the... ..."

"... In PAGE 5: ...work)case (i.e., a process instance), (iii) each event can have a performer also referred to as originator (the person executing or initiating the activity), and (iv) events have a timestamp and are totally ordered. Table1 shows an example of a log involving 19 events, 5 activities, and 6 originators. In addition to the information shown in this table, some event logs contain more information on the case itself, i.... In PAGE 5: ...Table 1: An event log. Event logs such as the one shown in Table1 are used as the starting point for mining. We distinguish three different perspectives: (1) the process perspective, (2) the organizational perspective and (3) the case perspective.... In PAGE 6: ... To illustrate the first two consider Figure 1. The log shown in Table1 contains information about five cases (i.e.... In PAGE 7: ...igure 1(a) does not show any information about the organization, i.e., it does not use any information on the people executing activities. However, Table1 shows information about the performers. For example, we can deduce that activity A is executed by either John or Sue, activity B is executed by John, Sue, Mike or Carol, C is executed by John, Sue, Mike or Carol, D is executed by Pete or Clare, and E is executed by Clare.... In PAGE 7: ...rom one individual to another, i.e., not focus on the relation between the process and individuals but on relations among individuals (or groups of individuals). Consider for example Table1 . Although Carol and Mike can execute the same activities (B and C), Mike is always working with John (cases 1 and 2) and Carol is always working with Sue (cases 3 and 4).... In PAGE 7: ... Although Carol and Mike can execute the same activities (B and C), Mike is always working with John (cases 1 and 2) and Carol is always working with Sue (cases 3 and 4). Probably Carol and Mike have the same role but based on the small sample shown in Table1 it seems that John is not working with Carol and Sue is not working with Carol. (Clearly the number of events in Table 1 is too small to establish these assumptions accurately.... In PAGE 7: ... Probably Carol and Mike have the same role but based on the small sample shown in Table 1 it seems that John is not working with Carol and Sue is not working with Carol. (Clearly the number of events in Table1 is too small to establish these assumptions accurately. However, for the sake of argument we assume that the things that did not happen will never happen.... ..."

"... In PAGE 4: ...efers to a case (i.e., a process instance), (iii) each event can have a performer also referred to as originator (the person executing or initiating the activity), and (iv) events have a timestamp and are totally ordered. Table1 shows an ex- ample of a log involving 19 events, 5 activities, and 6 originators. In addition to the information shown in this table, some event logs contain more information on the case itself, i.... In PAGE 4: ...Event logs such as the one shown in Table1 are used as the starting point for mining. We distinguish three difierent perspectives: (1) the process perspec-... In PAGE 5: ... To illustrate the flrst two perspectives consider Figure 1. The log shown in Table1 contains information about flve cases (i.... In PAGE 6: .... If activity B is executed, then also activity C is executed. However, for some cases activity C is executed before activity B. Based on the information shown in Table1 and by making some assumptions about the completeness of the log (i.e.... In PAGE 6: ...ther, i.e., not focusing on the relation between the process and individuals but on relations among individuals (or groups of individuals). Consider for example Table1 . Although Carol and Mike can execute the same activities (B and C), Mike is always working with John (cases 1 and 2) and Carol is always working with Sue (cases 3 and 4).... In PAGE 6: ... Although Carol and Mike can execute the same activities (B and C), Mike is always working with John (cases 1 and 2) and Carol is always working with Sue (cases 3 and 4). Probably Carol and Mike have the same role but based on the small sample shown in Table1 it seems that John is not working with Carol and Sue is not working with Mike.1 These examples show that the event log can be used to derive relations between performers of activities, thus result- ing in a sociogram.... In PAGE 6: ... For example, it is possible to generate a sociogram based on the transfers of work from one individual to another as is shown in Figure 1(c). Each node represents one of the six performers and each arc represents that there 1 Clearly the number of events in Table1 is too small to establish these assumptions accurately. However, for the sake of argument we assume that the things that did not happen will never happen.... In PAGE 7: ... Figure 1 does not address this. In fact, focusing on the case perspec- tive is most interesting when also data elements are logged but these are not listed in Table1 . The case perspective looks at the case as a whole and tries to establish relations between the various properties of a case.... In PAGE 9: ...e., one line in a table like Table1 . Each AuditTrailEntry element may contain Work owModelElement, EventType, Timestamp, and Originator elements.... In PAGE 9: ...e used to record the type of event (e.g., the start or completion of an activity or some exceptional behavior like the cancellation of a case). Table1 does not show any event types. However, one can always use the default event type com- plete.... In PAGE 12: ...g., in Table1 we only need to consider the flrst two columns. To construct a process model like in Figure 1(a), we need to be able to discover causal dependencies and decide on the types of splits and joins.... ..."

"... In PAGE 26: ...such as decision trees. To illustrate the idea of process mining consider the log shown in Table1 . Such a log could have been obtained by monitoring the SOAP messages the shipper service in Figure 3 exchanges with it its environment.... ..."

"... In PAGE 35: ... However, there are also techniques to discover organization models, social networks, and more data-oriented models such as decision trees. To illustrate the idea of process mining consider the log shown in Table5 . Such a log could have been obtained by monitoring the SOAP messages the shipper service in Figure 3 exchanges with it its environment.... In PAGE 35: ...3 They are all able to discover the shipper service as specified in Figure 3. Note that Figure 14 shows the names of the messages rather than the activities because this is the information shown in Table5 . Note that the algorithms used in Figure 14 can 3 Note that ProM allows for the mapping from one format to the other if needed.... In PAGE 36: ... However, ProM is not limited to DecSerFlow can can also be used to check the conformance of a specification in terms of abstract BPEL, EPC, or Petri nets. To illustrate this, assume that we add an additional process instance to Table5 where the notification is sent before the book is shipped to the customer (i.... ..."

"... In PAGE 7: ...Table 1: An example of an event log As we will show later, logs having a structure similar to the one shown in Table1 are recorded by a wide variety of CSCW systems. This information can be used to extract knowledge.... In PAGE 8: ...A B D E C F G H Figure 3: A process model derived from Table1 and represented in terms of a Petri net It is important to note that the Alpha algorithm is just one of the many process mining techniques available. For example, it is possible to extract a social network based on an event log.... In PAGE 10: ...he process perspective is concerned with the control-flow, i.e., the causal ordering of activities. Consider again Table1 . For the process perspective only the first two columns are relevant and the goal is to derive a process model, e.... In PAGE 11: ... The focus of mining this perspective is on discovering organizational structures and social networks. Note that Figure 3 completely ignores the third column in Table1 . Nevertheless this column may be used to derive interesting knowledge.... In PAGE 11: ... The data perspective is concerned with case and the data associated to cases. Table1 does not hold any data.... In PAGE 11: ...). Such information may be combined with the columns shown in Table1 to answer interesting questions such as: quot;Do large orders take more time than small orders? quot;, quot;What is the average flow time of cases where John is involved? quot;, quot;Does the treatment of male patients differ from the treatment of female patients? quot;. We have been working on techniques and tools to mine each of the three perspectives mentioned.... In PAGE 12: ... Instead of elaborating on these plug-ins we show some results based on the log shown in Table 1. Figure 6 shows the result of applying the Alpha algorithm [6] to the event log shown in Table1 . Note that indeed the process shown in Figure 3 is discovered.... In PAGE 12: ... Since ProM is multi-format it is also possible to represent processes in terms of an EPC or any other format added to the framework. Figure 6: Applying the Alpha plug-in to Table 1 Figure 7 shows a social network [3] based on the event log shown in Table1 . Now nodes represent actors rather than activities.... In PAGE 14: ... Figure 8: Screenshot of Staffware designer Figure 9 shows a fragment of a Staffware log. Note that the content of the log is similar to the content of the event log shown in Table1 . Therefore, process mining tools such as ProM have no problems using Staffware logs as input for process mining activities.... ..."

"... In PAGE 1: ... This log can have lots of data, such as the tasks that are executed, their time of execution, the person/system that performed them, the data elds related to these tasks, and so forth. For instance, consider the event log in Table1 . This log has four executions (cases) of a process that handles nes... In PAGE 3: ... If the log provides the tasks that are executed in the process and it is possible to infer their order of execution, the control- ow perspec- tive can be mined. The log in Table1 has this data (cf. elds \Case ID quot;, \Task Name quot; and \Timestamp quot;).... In PAGE 3: ... The organizational perspective dis- covers information like the social network in a process, the transferring of work etc. For instance, the log in Table1 shows that \Anne quot; transfers work for both \Mary quot; (case 2) and \John quot; (cases 3 and 4), and \John quot; sometimes transfers work for \Mary quot; (case 4). Besides, by inspecting the log, the mining algorithm could discover that \Mary quot; never has to send a reminder more than once, while \John quot; does not seem to perform as good.... ..."

"... In PAGE 6: ...Table 1: An example of an event log As we will show later, logs having a structure similar to the one shown in Table1 are recorded by a wide variety of systems. This information can be used to extract knowledge.... In PAGE 7: ...B D E C F G H Figure 2: A process model derived from Table1 and represented in terms of a Petri net. It is important to note that the Alpha algorithm is just one of the many process mining techniques available.... In PAGE 10: ...Consider again Table1 . For the process perspective only the first two columns are relevant and the goal is to derive a process model, e.... In PAGE 10: ... The focus of mining this perspective is on discovering organizational structures and social networks. Note that Figure 2 completely ignores the third column in Table1 . Nevertheless this column may be used to derive interesting knowledge.... In PAGE 10: ... The data perspective is concerned with case and the data associated to cases. Table1 does not hold any data.... In PAGE 10: ...). Such information may be combined with the columns shown in Table1 to answer interesting questions such as: quot;Do large orders take more time ... In PAGE 12: ...line in a table like Table1 . Each AuditTrailEntry element may contain WorkflowModelElement, EventType, Timestamp, and Originator elements.... In PAGE 12: ...ecord the type of event (e.g., the start or completion of an activity or some exceptional behavior like the cancellation of a case). Table1 does not show any event types. However, one can always use the default event type complete.... In PAGE 16: ...Table1 . Note that indeed the process shown in Figure 2 is discovered.... In PAGE 16: ... Since ProM is multi-format it is also possible to represent processes in terms of an EPC or any other format added to the framework. Figure 6: Applying the Alpha plug-in to Table 1 Figure 7 shows a social network based on the event log shown in Table1 . Now nodes represent actors rather than activities [2].... ..."

"... In PAGE 2: ....e. how to generate a model describing a process while only looking at event logs. Event logs such as the one shown in Table1 are used as the starting point for process mining, and from a process perspective we focus on the control- ow, i.... In PAGE 10: ... 4. The 5 cases of Table1 as transition systems. so trivial.... In PAGE 17: ...the iterative algorithm described in Section 5. Figure 7 shows a screenshot of ProM, clearly showing the Petri net of Figure 6 that was derived from the process log of Table1 using the region miner. Note that the plugin automatically inserts a common rst step, which is shown as a black transition in Figure 7.... ..."