Agile Long-term Workflows

Size: px

Start display at page:

Download "Agile Long-term Workflows"

Gwen Snow
10 years ago
Views:

1 Agile Long-term Workflows Dr. Mirjam Minor Daniel Schmalen Thomas Sauer Alexander Tartakovski Herculano de Biasi Prof. Dr. Ralph Bergmann -I -1-

2 We could not employ a workflow system that is not adaptable to changes. S. Rackow chip design expert from Silicon Image GmbH, Hannover -I -2-

3 Motivation Real world scenarios: Chip design Dynamics of technology, changes of the market Software development Changing customer requirements Healthcare Side-effects and other complications during the treatment of patients Major deviations from the usual business processes at run time Our approach: Agile workflows! -I -3-

Side-effects and other complications during the treatment of patients Major

4 Agile workflows For incremental and flexible modelling of processes at build-time + run-time: Workflow instances are derived from workflow definitions (WFD s, templates) Adaptation of ongoing processes by 1. Ad-hoc-changes of individual workflow instances e.g. [Reichert et al. 2003, Weber et al. 2005, Bassil et al. 2004] 2. Modifications to a workflow definition that is already in use by instances e.g. [Casati et al. 1998, Weske et al. 1999, Reichert et al. 2003] 3. Late-planning and hierarchical decomposition e.g. [van Elst et al. 2003, Freßmann et al. 2005] Our work fits in the first and third classifications. -I -4-

2005, Bassil et al. 2004] 2. Modifications to a workflow definition that is already in use by instances e.g. [Casati et al. 1998, Weske et al.

5 Top level WFD for chip design projects following the design flow SciWay 2.0 Acquisition Negotiation Project planning Design flow Project acceptance Maintenance Design flow Customer requirements specification Dummy design unit Top-level project execution Project context P3.0 P4

6 Sample top level instance Acquisition Negotiation Project planning Design flow Project acceptance Maintenance Design flow MC unit Project context Customer requirements specification P3.0 ME unit Top-level project execution P4

Maintenance Design flow MC unit Project context Customer

7 Sample instance for design unit MC [Review not ok] Preliminary FPGA synthesis Specification (unit specific part) [L HDL implementation Preliminary ASIC synthesis Unit specific review HDL implementation L] [Review ok] P3.1 P3.2 P3.3 STOP Unit context Check whether feature set confirmed Update implementation specification HDL implementation in addition P3.5 ASIC synthesis

HDL implementation L] [Review ok] P3.1 P3.2 P3.

8 Workflow modelling language Process modelling language is based on the notation of workflow patterns in [van Aalst et al. 2003]: Control flow elements 5 basic elements: Sequence, AND-split, AND-join, XOR-split, XOR-join LOOP-join, LOOP-split for structured cycles with one entry point and one exit point Breakpoints for reasons of adaptability Placeholder tasks for sub-workflows Further workflow elements Tasks, start symbols, end symbols Milestones Placeholder tasks for sub-diagrams modelling and monitoring only Context modelling language for influence factors is ontology-based -I -8-

one entry point and one exit point Breakpoints for reasons of adaptability Placeholder tasks for sub-workflows Further workflow elements Tasks,

9 Tool Demonstration -I -9-

10 Case-based authoring support Basic idea of case-based reasoning (CBR): Past problem Past solution similar useful? Current problem Current solution? Employment of CBR to workflow modeling: Reuse of experience knowledge how to adapt workflow instances Development of a suitable representation and similarity assessment for workflow instances -I -10-

Employment of CBR to workflow modeling: Reuse of experience knowledge how to adapt

11 Similarity measure for workflow instances 1. Context: conventional similarity function according to the local-global principle, e.g. weighted sum 2. Control flow: similarity function based on graph edit distance, e. g. [Bunke, Messmer, 1993]: x, y be directed, labeled graphs, o s be atomic edit operations (insert node, insert edge, delete node, delete edge, change node label, change edge label), c(o) [0,1] be a cost function: δ ( x, y) k = min{ c( o i= 1 i ) ( o 1,..., ok ) transforms x to y} Overall similarity of two cases: aggregation of the values of 1. and 2. -I -11-

12 Overview research topics/projects CAKE: Collaborative, agile knowledge engine Agile long-term workflows URANOS BMBF project Authoring support / change reuse Collaboration patterns AMIRA EU project Workflow enactment tracking rjm business solutions Reference models for agile workflow systems (partial) transformation of agile workflows to and from BPEL Medical workflows cooperation with Uniklinik Greifswald -I -12-

enactment tracking rjm business solutions Reference models for agile workflow systems (partial)

13 Thank you! Questions? Remarks? -I -13-

BIS 3106: Business Process Management. Lecture Two: Modelling the Control-flow Perspective

BIS 3106: Business Process Management Lecture Two: Modelling the Control-flow Perspective Makerere University School of Computing and Informatics Technology Department of Computer Science SEM I 2015/2016