Jan Mendling and Stefanie Rinderle-Ma, eds.: Proceedings of EMISA 2016,
Gesellschaft für Informatik, Bonn 2016
Discovering Interacting Artifacts from ERP Systems
(Extended Abstract)3
D. Fahland1, X. Lu1, Marijn Nagelkerke2, Dennis van de Wiel2
Abstract: Enterprise Resource Planning (ERP) systems are widely used to manage business doc-
uments along a business processes and allow very detailed recording of event data of past process
executions and involved documents. This recorded event data is the basis for auditing and detecting
unusual flows. Process mining techniques can analyze event data of processes stored in linear event
logs to discover a process model that reveals unusual executions. Existing techniques assume a linear
event log that use a single case identifier to which all behavior can be related. However, in ERP sys-
tems processes such as Order to Cash operate on multiple interrelated business objects, each having
their own case identifier, their own behavior, and interact with each other. Forcing these into a single
case creates ambiguous dependencies caused by data convergence and divergence which obscures
unusual flows in the resulting process model. We present a new semi-automatic, end-to-end approach
for analyzing event data in a plain database of an ERP system for unusual executions. We identify an
artifact-centric process model describing the business objects, their life-cycles, and how the various
objects interact along their life-cycles. The technique was validated in two case studies and reliably
revealed unusual flows later confirmed by domain experts. The work summarized in this extended
abstract has been published in [Lu15].
Keywords: Process Mining, ERP-System, Artifact-Centric Model, Object Life-Cycle, Interaction
Discovery
1 Introduction and Problem Description
Information systems (IS) not only store and process data in an organization but also record
event data about how and when information changed. This “historical event data” can be
used to analyze, for instance, whether information processing in the past conformed to the
prescribed processes or to compliance requirements. Process mining [Aa11] offers auto-
mated techniques for this task. In particular exploring visual models discovered from event
data allows to identify unusual flows and their circumstances; based on which concrete
measures for process improvement can be devised [Ec15]. Prerequisite to this analysis is
a process event log that holds events about information changes with the assumption that
each event belong to one specific execution of a specific process.
In general, information access is not tied to a particular process execution; rather the same
information can be accessed and changed from various processes and applications. For
1 d.fahland@tue.nl, Eindhoven University of Technology, The Netherlands
2 KPMG IT Advisory N.V., Eindhoven, The Netherlands
3 This article summarizes problem, approach, and selected findings of a study published as Xixi Lu, Marijn
Nagelkerke, Dennis van de Wiel, and Dirk Fahland. Discovering Interacting Artifacts from ERP Systems. Ser-
vices Computing, IEEE Transactions on, 8(6), 2015 doi:10.1109/TSC.2015.2474358 [Lu15].
1
F2
Sales documents (SD) Delivery documents (DD)
SD id Date created Reference id Document type Value Last change DD id Date created Reference SD id Reference BD Document type Picking date
S1 16-5-2020 null Sales Order 100 10-6-2020 D1 18-5-2020 S1 B1 Delivery 31-5-2020
S2 17-5-2020 null Sales Order 200 31-5-2020 D2 22-5-2020 S1 B2 Delivery 5-6-2020
S3 10-6-2020 S1 Return Order 10 NULL D3 25-5-2020 S2 B2 Delivery 5-6-2020
F1 D4 12-6-2020 S3 null Return Delivery NULL
F3
F4
Documents Changes Billing documents (BD)
Change id Date changed Reference id Table name Change type Old Value New Value BD id Date created Document type Clearing date
1 17-5-2020 S1 SD Price updated 100 80 B1 20-5-2020 Invoice 31-5-2020
2 19-5-2020 S1 SD Delivery block released X - B2 24-5-2020 Invoice 5-6-2020
3 19-5-2020 S1 SD Billing block released X -
4 10-6-2020 B1 BD Invoice date updated 20-6-2020 21-6-2020 Parent Child
table table
Fig. 1: The tables of the simplified OTC example
(Sales Order) Sales order
Divergence 7 events “Created” related to S1 Sales order created
(a) (Delivery)
2 1 Return Return
1 order 1 delivery
(Invoice) (Return Order) Delivery 2 Invoice
1
(Delivery) created created created created
3 3 1 1
2
(Invoice) (Return Delivery) (b)
Delivery 1 Invoice Legend:
created created artifact
Sales order 2
3 2 2 Event type
created
2 Return order Return delivery Causal relation
1 or interaction
(Sales Order) (Invoice) 1 created created
(Delivery)
1 1 Deviating
3 events “Created” related to S2 Convergence (c) interaction
Fig. 2: Creation of documents of Fig. 1 along time (a), a classic process model based on a single case
identifier (b), and an artifact-centric model based on 5 case identifiers (c).
instance, in Enterprise Resource Planning (ERP) systems (such as SAP and Oracle En-
terprise), information is stored in business objects (or documents) which are linked via
one-to-many and many-to-many relations, typically in the relational database. The objects
themselves are encapsulated in services [AMZ00] which are invoked by high-level end-
to-end business processes; each invocation is called a transactions which is logged in the
data object itself.
Fig. 1 shows a simplified example of the transactional data of an Order to Cash (OTC)
process supported by SAP systems; Fig. 2(a) visualizes the events of Fig. 1 that are related
to document creation. There are two sales orders S1 and S2; creation of S1 is followed
by creation of a delivery document D1, an invoice B1, another delivery document D2, and
another invoice B2 which also contains billing information about S2. Creation of S2 is
also followed by creation of another delivery document D3. Further, there is a return order
S3 related to S1 with its own return delivery document D4. The many-to-many relations
between documents surface in the transactional data of Fig. 1: a sales document can be
related to multiple billing documents (S1 is related to B1 and B2) and a billing document
can be related to multiple sales document (B2 is related to S1 and S2). This behavior
already contains an unusual flow: delivery documents were created twice before the billing
document (main flow), but once the order was reversed (B2 before D3).
When applying classical process mining techniques, one first has to extract an event log
based on a single case identifier to which all event data can be related. Choosing SD id
in Fig. 1 leads to the two sequences of events shown in Fig. 2(a). Process discovery on
2
this log yields the model of Fig. 2(b) which is wrong: two invoices are created before
their deliveries instead of one, and three invoices are created instead of two (known as
divergence and convergence, respectively) [Pi11].
2 Approach: Discovering Artifact-Centric Models
We propose to approach the problem under the “conceptual lens” of artifact-centric mod-
els [CH09]. An artifact is a data object over an information model; each artifact instance
exposes services that allow changing its informational contents; a life-cycle model gov-
erns when which service of the artifact can be invoked; the invocation of a service in one
artifact may trigger the invocation of another service in another artifact. Information mod-
els of different artifacts can be in one-to-many and many-to-many relations allowing to
describe behavior over complex data in terms of multiple objects interacting via service
invocations. Under this lens, each document of an ERP system can be seen as an arti-
fact; a transaction on a document is a service call on the artifact; behavioral dependencies
between transactions of documents can be seen as life-cycle behavior and dependencies
of service calls. Describing the transactional data of Fig. 1 with artifact-centric concepts
yields the model of Fig. 2(c); it visualizes the order in which objects are created and also
highlights the unusual flow of invoice B2 being created before delivery D2.
The problem of discovering an artifact-centric
Data Database
process model from relational ERP data de- Source Schema
composes into two sub-problems. (1) Given a
relational data source, identify a set of artifacts,
extract for each artifact an event log, and dis- 1.1 Discover
2.1 Discover Artifact
cover a model of its life-cycle. (2) Given a set Type
of artifacts and their data source, identify inter- Type-Level
actions between the artifacts, between their in- Interaction
2.2 Add case
stances, between their event types and between 1.2 Extract
references
their events. Figure 3 shows the overview of
Case A1 Case B1 Case C1
our approach. (1.1) We use the data schema of Case A2 Case B2 Case C2
the data source to discover artifact types which Case An Case Bm Case Ck
detail all timestamped columns related to a par- Event
2.3 Discover
ticular business object. (1.2) For each artifact 1.3 Discover Log
we then extract a classical event log [Aa11],
each case describes all events related to one
instance of the artifact. (1.3) Existing process
discovery algorithms allow discovering a life-
cycle model of the artifact. In parallel, (2.1) Life-Cycle
Models... + Activity-Level Interactions
we discover interactions between artifacts from
foreign key relations in the data source; (2.2) Fig. 3: An overview on our approach.
during log extraction, each case of an artifact is
annotated with references to cases of other artifacts this case interacts with. (2.3) The case
references are refined into interactions between activities of different artifact life-cycles.
3
3 Results
We implemented our approach based on [NvDF12] and conducted two case studies. By
separating data into artifacts along one-to-many relations, we eliminated divergence and
convergence, the interaction flows discovered from one-to-many relations were meaning-
ful to business users, and unusual flows were detected.
Fig. 4 shows models obtained from 2 1140
Created
months data of an SAP Order-to-Cash pro-
646
2581
1472 907 39
Delivery H_Created 2439 55 31 6
5118
12 23 65 346
597 47 Payment05or15_Payment Received 107
822
278 26 46
641 Invoice H_Created 142 49 341
cess (11 document header tables, 134,826
2629
53 371 620 42 1 14
46 316 DebitMemoRequest H_Created 354 78 59 18
2
1360 1298 2 212 141
1 1 15 1 DebitMemo H_Created 3 13 1 1
14
1 5 9 90 138 1 304
271 1 1 3 PostInAR_PostedInAR 854 1 17 10
records of 5-49 attributes); the model at the
3479
22 1 1 1 1 568
625 Contract H_Created 89 108
741
14 8 ReturnDelivery H_Created 9 22 18
1
15 15 6 ProFormaInvoice H_Created 257
730
1 734 2 2
top was obtained with a classical approach
8 CreditMemoRequest H_Created 6 2 InvoiceCancellation H_Created 2
33 40
7 12 49
5 CreditMemo H_Created 2 4
34
1 7
ReturnOrder H_Created
1
(only 29 of 77 edges are correct); the
model at the bottom was obtained using
our approach. In both case studies the dis-
covered process models were assessed as
accurate graphical representations of the
source data by domain experts; all edges
including outlier edges were assessed as
Fig. 4: SAP OTC process, classical process model
correct and traced back to the source data
obtained from single event log (top) and artifact-
together with domain experts. These in- centric model highlighting outliers (bottom)
sights could be obtained exploratively and
much faster than with existing best practices.
References
[Aa11] Aalst, W.M.P. van der: Process Mining: Discovery, Conformance and Enhancement of
Business Processes. Springer, 2011.
[AMZ00] Al-Mashari, Majed; Zairi, Mohamed: Supply-chain re-engineering using enterprise re-
source planning (ERP) systems: an analysis of a SAP R/3 implementation case. IJPDLM,
30(3/4):296–313, 2000.
[CH09] Cohn, D.; Hull, R.: Business artifacts: A data-centric approach to modeling business
operations and processes. Bulletin of the IEEE Computer Society TCDE, 32(3):3–9,
2009.
[Ec15] van Eck, Maikel L.; Lu, Xixi; Leemans, Sander J. J.; van der Aalst, Wil M. P.: PM ˆ2
: A Process Mining Project Methodology. In: CAiSE 2015. volume 9097 of LNCS.
Springer, pp. 297–313, 2015.
[Lu15] Lu, Xixi; Nagelkerke, Marijn; van de Wiel, Dennis; Fahland, Dirk: Discovering Inter-
acting Artifacts from ERP Systems. IEEE Trans. Services Computing, 8(6):861–873,
2015.
[NvDF12] Nooijen, Erik H. J.; van Dongen, Boudewijn F.; Fahland, Dirk: Automatic Discovery
of Data-Centric and Artifact-Centric Processes. In: DAB’12. volume 132 of LNBIP.
Springer, pp. 316–327, 2012.
[Pi11] Piessens, D.A.M.: Event Log Extraction from SAP ECC 6.0. Master’s thesis, Eindhoven
University of Technology, 2011.
4