Andreas Martin - FHNW A Case Modelling Language for Process Variant Management in Case-based Reasoning Riccardo Cognini 2, Knut Hinkelmann 1 and Andreas Martin 1 AdaptiveCM 2015 – 31.08.2015 4th International Workshop on Adaptive Case Management and other non-workflow approaches to BPM 1 FHNW University of Applied Sciences and Arts Northwestern Switzerland, School of Business, Olten, Switzerland 2 University of Camerino, School of Science and Technology, Computer Science Department, Camerino, Italy A Case Modelling Language for Process Variant Management in Case-based Reasoning 1 

Andreas Martin - FHNW Introduction Problem / Research objective A Case Modelling Language for Process Variant Management in Case-based Reasoning 2 

Andreas Martin - FHNW Problem Knowledge work is not routine work …  “[...] the sequence of actions depends so much upon the specifics of the situation [...] necessitating that part of doing the work is to make the plan itself" (Swenson, Palmer and Silver, 2011, p.8) .  Difficult to predict upcoming tasks.  Hard to determine the type and scope of tasks.  Sequence of tasks may vary due to already achieved results and unforeseeable events. A Case Modelling Language for Process Variant Management in Case-based Reasoning 3 

Andreas Martin - FHNW Problem  Situation: As the term knowledge work implies, knowledge worker need experience and (procedural) knowledge to find a solution (approach, plan, activities) to specific situation / problem.  Underlying assumption: For knowledge work it seems useful to take approaches that structure business processes just in part as process fragments.  “Process fragments are reflecting the partial and intermittent knowledge one modeller [or a knowledge worker] has at a certain time about a specific situation" (Eberle, Unger and Leymann, 2009, p.399) . A Case Modelling Language for Process Variant Management in Case-based Reasoning 4 

Andreas Martin - FHNW Research objective  Knowledge workers are required to make decisions based on process fragments, which can only be made by the knowledge workers themselves at run-time.  Research objective: Therefore, a vocabulary and content representation (modelling language) is needed that support the manual planning, modelling, generation and refinement of process fragments during run-time.  Underlying method: We use case-based reasoning (CBR), to put the experience management into the hands of the knowledge worker and provide a way to retrieve and reuse historic process fragments. A Case Modelling Language for Process Variant Management in Case-based Reasoning 5 

Andreas Martin - FHNW What is Case-based Reasoning (CBR)?  CBR can be seen as “reasoning by remembering”…  and it is a technically independent methodology to humans and information systems.  “Case-based reasoning is both […] the ways people use cases to solve problems and the ways we can make machines use them”.  CBR has been applied in business process contexts, e.g., for workflow retrieval, adaptation, construction and monitoring. A Case Modelling Language for Process Variant Management in Case-based Reasoning 6 

Andreas Martin - FHNW What is Case-based Reasoning (CBR)? CBR- CYCLE (& CASE- BASE)  Retrieve the most similar cases from the case-base containing previous cases, based on the problem description of the new case using a similarity mechanism.  Reuse the knowledge in the retrieved case(s) in order to solve the current problem – adapt the historical knowledge to the new problem (adaptation).  Revise and test the suggested solution e.g. by evaluating it under the real world situation (evaluation).  Retain useful experience for future reuse and store a new case in the knowledge base (case learning). A Case Modelling Language for Process Variant Management in Case-based Reasoning 7 Based on: A. Aamodt and E. Plaza, “Case-Based Reasoning : Foundational Issues , Methodological Variations , and System Approaches,” Artificial Intelligence Communications, vol. 7, no. 1, pp. 39–59, 1994. 

Andreas Martin - FHNW Application Scenario A Case Modelling Language for Process Variant Management in Case-based Reasoning 8 

Andreas Martin - FHNW Master Study Admission Process A Case Modelling Language for Process Variant Management in Case-based Reasoning 9 

Andreas Martin - FHNW (Sub-) Tasks based working Used for data collection  The end user are familiar with an task / task pattern concept.  The data used in this approach was initially gathered in an EU project called MATURE: A Case Modelling Language for Process Variant Management in Case-based Reasoning 10 

Andreas Martin - FHNW CASE CASE Demanding*) example 1. In the first case the problem is solved by calling the university of the applicant student asking for some information. 2. In the second case the university is called and then proof of the existence of the university. 3. In the last case just the proof of the of existence of the university is requested. *) These three examples are not the demanding “thing”, it is the fact that an uncountable and unknown variation of these process fragments can exist. A Case Modelling Language for Process Variant Management in Case-based Reasoning 11 CASE 

Andreas Martin - FHNW Application Scenario Analysis  Application Scenario Analysis: Knowledge workers (as shown in application scenario) currently working with concrete process fragments / (sub-) tasks.  As improvement: Case generalization and abstraction can reduce the complexity of the cases, increase the flexibility and reduce the size of the case base to enhance the retrieval efficiency.  Therefore we introduce a case-based process fragment modelling language that supports the manual generation and refinement of generalized cases in CBR based approach. A Case Modelling Language for Process Variant Management in Case-based Reasoning 12 

Andreas Martin - FHNW Approach Comparison A Case Modelling Language for Process Variant Management in Case-based Reasoning 13 

Andreas Martin - FHNW Drawbacks of: BPMN and other Imperative Languages End users …  do not have enough time, knowledge and skills to model or update a BPMN model.  are able to specify which activities should be performed and by whom,  but they are not able to establish a temporal order,  because they are focused just on their own tasks. Imperative languages …  are designed to express something that is fully defined including all the possible aspects.  “is like modifying” a software source code written by someone else. A Case Modelling Language for Process Variant Management in Case-based Reasoning 14 

Andreas Martin - FHNW CMMN, a declarative language  is a designed to model non predefined, partially structured and non repeatable business processes.  Mandatory and optional activities can be modelled without  specifying an execution order or  the situation in which an activity can be executed. Issue of CMMN …  it is not possible to specify complex execution criteria.  E.g. it is not possible to specify at least one of the activities in a set has to be executed.  complex data elements are not provided;  it implies that just little information about the type of data or document is available. A Case Modelling Language for Process Variant Management in Case-based Reasoning 15 

Andreas Martin - FHNW Comparison of Modelling Languages  We propose the use of BPFM notation as a language for case representation in CBR. BPFM:  permits defining the BP activities that must or can be performed with and without including an execution order  considers complex constraints and different types of data objects.  is a configurable process model since it can encapsulate more than one BP variant A Case Modelling Language for Process Variant Management in Case-based Reasoning 16 

Andreas Martin - FHNW The Approach… …Business Process Feature Model Notation A Case Modelling Language for Process Variant Management in Case-based Reasoning 17 

Andreas Martin - FHNW Business Process Feature Model (BPFM)  The Business Process Feature Model is constituted by a tree of related activities.  The root identifies the main services.  Each internal (non-leaf) activity denotes a sub-process that can be further refined.  The external (leaf) activity represents an atomic task.  BPFM allows using the same meaning and graphical representation given by BPMN 2.0.  A BPFM model allows for the defining constraints between activities. A Case Modelling Language for Process Variant Management in Case-based Reasoning 18 

Andreas Martin - FHNW Mapping BPFM Activities to BPMN A Case Modelling Language for Process Variant Management in Case-based Reasoning 19 

Andreas Martin - FHNW Constraints in Business Process Feature Model (BPFM) Constraints are used to express  if child activities can or have to be selected in the configuration to be included in the BP variant, and  if they can or have to be included in each execution path of the BP variant. A Case Modelling Language for Process Variant Management in Case-based Reasoning 20 

Andreas Martin - FHNW Mandatory Constraints A Case Modelling Language for Process Variant Management in Case-based Reasoning 21  Mandatory Constraint requires that the connected child activity must be inserted in each BP variant,  and it must be included in each execution path. 

Andreas Martin - FHNW Special Case Constraints A Case Modelling Language for Process Variant Management in Case-based Reasoning 24  A Special Case Constraint allows for the connected child activity to be inserted in each BP variant.  When it is inserted it must be included in each execution path. 

Andreas Martin - FHNW Inclusive Constraints A Case Modelling Language for Process Variant Management in Case-based Reasoning 25  An Inclusive Constraint requires that at least one of the connected child activities be inserted in each BP variant,  and at least one of them must be included in each execution path. 

Andreas Martin - FHNW One Selection Constraints A Case Modelling Language for Process Variant Management in Case-based Reasoning 27  A One Selection Constraint requires that exactly one of the connected child activities be inserted in each BP variant,  and it must be included in each execution path. 

Andreas Martin - FHNW Data Objects in BPFM A Case Modelling Language for Process Variant Management in Case-based Reasoning 30  BPFM manages all types of BPMN 2.0 data objects, including data object states, with the same modelling notation 

Andreas Martin - FHNW Data Mapping to BPMN A Case Modelling Language for Process Variant Management in Case-based Reasoning 31 

Andreas Martin - FHNW User Acceptance? A Case Modelling Language for Process Variant Management in Case-based Reasoning 32 

Andreas Martin - FHNW User Acceptance? A Case Modelling Language for Process Variant Management in Case-based Reasoning 33 

Andreas Martin - FHNW User Acceptance? A Case Modelling Language for Process Variant Management in Case-based Reasoning 34 

Andreas Martin - FHNW The Approach… …BPFM in Case-based Reasoning (CBR) A Case Modelling Language for Process Variant Management in Case-based Reasoning 35 

Andreas Martin - FHNW Back to Case-based Reasoning (CBR) THE CASE  Traditional CBR terminology: a case consists of a problem space that is used for describing a certain solution space.  Our CBR terminology: a case consists of a case characterization (derived from Bergmann) that is used for describing a certain case content. A Case Modelling Language for Process Variant Management in Case-based Reasoning 36 Characterization Content CASE 

Andreas Martin - FHNW Retrieval Description Case-based Reasoning Cycle - Retrieval A Case Modelling Language for Process Variant Management in Case-based Reasoning 38 Retrieve CASE BASE NEW Situation How is: the configuration (model) and the retrieval realized ? 

Andreas Martin - FHNW  ArchiMEO is an enterprise ontology based on ArchiMate and is extended with selected concepts from other enterprise ontologies.  ArchiMEO is implemented using RDF(s) and OWL.  ArchiMEO has been developed by several team members of the FHNW Information and Knowledge Management Research Group (IKM).  ArchiMEO is licensed under a Creative Commons Attribution- ShareAlike 3.0 Unported License.  ArchiMEO is available for download as TTL- files (Terse RDF Triple Language) under: ikm-group.ch/archimeo A Case Modelling Language for Process Variant Management in Case-based Reasoning 39 ArchiMate is a technical standard from The Open Group. RDF(S) / OWL is a W3C standard. The enterprise ontology ArchiMEO is based on ArchiMate. 

Andreas Martin - FHNW Ontology-based Case-based Reasoning (OBCBR) A Case Modelling Language for Process Variant Management in Case-based Reasoning 40 

Andreas Martin - FHNW Case Characterization describing Process Knowledge A Case Modelling Language for Process Variant Management in Case-based Reasoning 41 Further elements:  TaskObjective: The task objective element describes the goal of the task itself. This is similar to the name and/or description of an BPMN activity.  TaskRole: The task role element is used to describe the role of the involved person of the task. Through the inclusion of an enterprise or domain ontology, it is possible to reuse an existing enterprise specific role / organizational model.  TaskUser: The task user elements is used to indicate the person who described the case. 

Andreas Martin - FHNW Demonstration A Case Modelling Language for Process Variant Management in Case-based Reasoning 42 

Andreas Martin - FHNW Learning Retrieval Adaptation Description Extended Case-based Reasoning Cycle A Case Modelling Language for Process Variant Management in Case-based Reasoning 43 Reuse Retrieve CASE BASE Retain Revise / Generalize Revise NEW Situation Evaluation 

Andreas Martin - FHNW Concrete CASEs A Case Modelling Language for Process Variant Management in Case-based Reasoning 44 Characterization Content CASE A Characterization Content CASE B 

Andreas Martin - FHNW Generalized CASE A Case Modelling Language for Process Variant Management in Case-based Reasoning 45 Characterization Content Generalized CASE 

Andreas Martin - FHNW Implementation A Case Modelling Language for Process Variant Management in Case-based Reasoning 46  ICEBERG Toolkit  BPFM Modelling Platform 

Andreas Martin - FHNW Interlinked Case-based Reasoning (ICEBERG) A Case Modelling Language for Process Variant Management in Case-based Reasoning 47  ICEBERG Approach & Toolkit: Using interlinked (ontology- based) case-based reasoning to bring hidden knowledge to the surface. 

Andreas Martin - FHNW ICEBERG - System architecture  Technology:  Apache Jena: open source Java framework for building Semantic Web applications  OpenDolphin: open-source library for a lightweight remote model-view- controller separation.  JavaFX: GUI framework  TopBraid Composer: an ontology engineering software (paid) A Case Modelling Language for Process Variant Management in Case-based Reasoning 48 

Andreas Martin - FHNW Adapting case models / content BPFM Modelling Platform an OMiLAB Project A Case Modelling Language for Process Variant Management in Case-based Reasoning 49 

Andreas Martin - FHNW Conclusion and Future Work A Case Modelling Language for Process Variant Management in Case-based Reasoning 50 

Andreas Martin - FHNW Conclusion and Future Work  This paper presented an approach to model cases in knowledge-intensive BPs.  The approach merges CBR with BPFM notation in order to represent cases.  We applied the approach to a concrete case in a public administration scenario in order to show its suitability. Future Work: We plan to …  … make further evaluation with the respect to usability of different modelling languages / elements.  … transfer the approach to a different scenario. A Case Modelling Language for Process Variant Management in Case-based Reasoning 51