ENABLING END USERS TO PROACTIVELY TAILOR

UNDERSPECIFIED, HUMAN-CENTRIC BUSINESS PROCESSES

“Programming by Example” of Weakly-Structured Process Models

Todor Stoitsev, Stefan Scheidl

SAP Research, SAP AG, Bleichstr. 8, Darmstadt, Germany

Felix Flentge, Max Mühlhäuser

Telecooperation Group, Darmstadt University of Technology, Darmstadt, Germany

Keywords: End user development, computer supported cooperative work, ad-hoc workflow, knowledge management,

human computer interaction.

Abstract: Enterprises face the challenge of managing underspecified, human-centric business processes, which are

executed in distributed teams in a rather informal, ad-hoc manner. This gave hibernating CSCW and ad-hoc

workflow research a new push recently. However, there is still the need to clearly perceive end users as the

actual drivers of business processes and to enable them to proactively tailor these processes according to

their actual expertise and problem solving strategies. This paper presents the design and evaluation of a

prototype for end-user development of weakly-structured process models through email-integrated task

management. The presented CTM (Collaborative Task Manager) prototype uses “programming by

example” to leverage user experience with standard email and task management applications and to extend

user skills towards the definition of reusable process structures. By closely correlating to the actual user

work practices and software environment, the tool provides a “gentle slope of complexity” for end users

engaging in process tailoring activities.

1 INTRODUCTION

Up until recently, workflow systems were too formal

and restrictive to be useful for knowledge-intensive

and rather informal processes (Schwarz et al., 2001).

The importance of such processes and the increase

of distributed team work led to further research on

enterprise efficiency, which clearly presents how

“individual actions lead to overall enterprise

performance” (Wiig, 2004). It becomes apparent that

the traditional enterprise process modelling

perspective is being replaced by tailoring of business

processes according to the individual point of view

and connecting them towards the achievement of

common enterprise goals. This novel view on

business processes emerges in analyst reports as the

“Process of Me” (Gartner, 2006) and is recognized

as one of the major challenges for the next

generation Business Process Management (BPM). It

states the fundamental need to provide end users

with adequate techniques to proactively express

process knowledge and to participate in business

process management and design.

End User Development (EUD) is defined as “a

set of methods, techniques, and tools that allow

users of software systems, who are acting as non-

professional software developers, at some point to

create, modify, or extend a software artefact”

(Lieberman et al., 2006). Within the presented paper

a process model is considered as a software artefact,

which can be adapted and enacted to support

underspecified, human-centric processes. The

presented study is motivated through the possibility

to “render” appropriation of process models to end

users and to “exploit the potential of opportunity-

based and emergent changes” from the introduction

of groupware in enterprises (Wulf & Jarke, 2004).

Riss et al. (2005) discuss the challenges for the

next generation BPM by suggesting the recognition

and reuse of “task patterns” and “process patterns”

as alternative to static workflows. However,

concrete examples for engaging business users in

task pattern definition and modelling towards

Stoitsev T., Scheidl S., Flentge F. and Mühlhäuser M. (2008).

ENABLING END USERS TO PROACTIVELY TAILOR UNDERSPECIFIED, HUMAN-CENTRIC BUSINESS PROCESSES - “Programming by Example”

of Weakly-Structured Process Models.

In Proceedings of the Tenth International Conference on Enterprise Information Systems - HCI, pages 38-46

DOI: 10.5220/0001681300380046

 SciTePress

generic enterprise process models are still missing as

well as techniques for achieving that. This issue is in

the focus of the presented paper. The described

approach ensures a “gentle slope of complexity”

(MacLean et al., 1990) for process tailoring

activities by leveraging user experience with

standard tools for collaboration (email) and task

management (to-do lists) and extending user skills

towards definition of weakly-structured process

models through “programming by example”

(Lieberman, 2001). This EUD technique enables

unobtrusive support by embedding the process

definition in the existing end user working

environment and inferring process models from the

captured executed activities. The described approach

presents a valuable extension to “evolutionary”

workflows (Hermann, 2000) and “interactive

process models” (Jorgensen, 2004) by allowing

“seeding, evolutionary growth, and reseeding”

(Fisher et al., 2004) of weakly structured process

models in shared enterprise repositories and task

instance-based evolution tracking. The iterative,

evolutionary transitions from execution to design

(and vice versa) of adaptable, weakly-structured

process models exceed the capabilities of known

email-based workflows (Agostini et al., 1997).

In section 2 we present basic problems regarding

current practices in ad-hoc processes, which are used

to introduce process tailoring by end users. Section 3

presents a prototype for end user driven process

definition. Section 4 describes results from prototype

evaluation at a partner company. In section 5 we

give conclusions and future research directions.

2 ADRESSED PROBLEM AREAS

The presented study builds up on state of the art

research in the areas of task management, flexible

workflows, CSCW and EUD. It is based on intra-

organizational knowledge sources accumulating

customer requirements as well as on dedicated site

visits and interviews at three companies from

various industries: textile (120 employees), software

(ca. 500 employees), automotive (ca.

150

employees). Based on the preliminary studies we

identified five generic problem areas concerning

user work practices in ad-hoc processes that can be

used to introduce user-driven process composition:

Lacking Transparency. Email is the main tool for

exchange of tasks and task-related information in

informal processes (Bellotti et al., 2005). Users

further organize tasks in to-do lists (Bellotti et al.,

2004). These tools do not provide end-to-end

overview of running collaborative activities.

No Structured Storage and Retrieval of Process

Knowledge. Users spent considerable effort to

search for task-related data in email folders (Bellotti

et al., 2004). While having individual strategies for

storing data in email and file folders, users are not

able to predict how their “sorting” practice will scale

over time. Increasing data amount increases search

effort and user efficiency degrades.

Lacking Exchange of Process Knowledge. As

process knowledge often remains implicit, stuck in

personal email and file folders, people “know” what

to do but cannot share it efficiently with their

colleagues. This leads to problems when domain

experts are not available and cannot provide support

on time critical activities.

Disjunction between Best-practices and Running

Processes. A common way to store process

guidelines is in text documents (e.g. Microsoft

Word). Text representations do not provide the

possibility to follow evolving user tasks with respect

to the provided guidelines and to observe to what

extent the described (best) practice is being

followed, or why deviations have occurred.

Inability to Trace Evolving Best-practices. Best-

practices for informal processes may often change

due to the changing business conditions. Having

previous process information in email and file

folders and guidelines in text-based documents does

not allow structured comparison and reasonable

evaluation to what extent best-practices need to be

adapted or if different variations have to be managed

for different application contexts.

3 COLLABORATIVE TASK

MANAGER (CTM)

The Collaborative Task Manager (CTM) is an email-

integrated task management tool, with extensive

support for definition, adaptation and reuse of

weakly-structured process models. All industry

partner companies involved in our preliminary

studies were using Microsoft Outlook (OL) as a

standard email client. To ensure an integrated

support within the common working environment,

CTM is delivered as an OL Add-In, additionally

exploiting the fact that tasks and email are provided

in the same office application. The CTM Add-In

provides extensions of the OL mail and task items

and enables “programming by example”

(Lieberman, 2001) by using web services to track

user actions, executed on CTM tasks and replicating

ENABLING END USERS TO PROACTIVELY TAILOR UNDERSPECIFIED, HUMAN-CENTRIC BUSINESS

PROCESSES - “Programming by Example” of Weakly-Structured Process Models

Figure 1: CTM To-Do List (TDL).

data on a central server. The data is held in a

Database (DB) that provides a central tracking

repository for all CTM users. Tracking of email

communication for task delegation integrates the

individual task hierarchies of different users to

overall enterprise process structures, emerging on

the server. The CTM Add-In application provides

“Process Info” links on tasks and task-related email

messages, which open a web-based client, providing

overview and navigation in the generated process

structures by retrieving data from the server.

3.1 CTM To-Do List (TDL)

The TDL is shown in Figure 1. CTM extends OL

tasks with functionality for displaying a hierarchical

tree structure. The Add-In provides additional

toolbars for direct access to the main CTM

functionalities. CTM enables insertion and removal

of tasks and sub tasks in a task hierarchy in a light-

weight manner. Task insertion opens a new OL task

dialog where the user works with the familiar OL

task fields. Files can be added to CTM tasks as

common OL task attachments. An email can be

saved as CTM task, whereby the mail subject, body

and attachments are accordingly applied to the task.

3.2 Transfer of Tasks and Deliverables

A CTM task is delegated through a preformatted

“Request” message. Recipients can “Accept”,

“Decline” or “Negotiate” the request. While

request/accept/decline are standard actions known

also from the exchange of meeting requests in OL,

iterative negotiations allow additional clarifications

Figure 2: Detailed task dialog overview.

on tasks. The actual discourse takes place in the

email text, which is independent from the given

message type. This allows open-ended collaboration

on tasks and prevents from submitting user

behaviour to strict speech-act rules, which is a

known limitation in speech-acts adoption (Button,

1994). When a request is accepted, and later on

completed by a recipient, the latter issues a “Declare

Complete” message. Hereupon the requester can

respond with “Approve Completion” or “Decline

ICEIS 2008 - International Conference on Enterprise Information Systems

Completion” message. These additional actions

allow negotiation of deliverables, before the final

completion of a delegated task. To avoid flooding of

the OL inbox with task-related messages, a “Move

CTMs” button is provided which moves all task

related emails to a special CTM mail folder.

All email exchange related to a task is associated

to a task dialog and stored on the server. Dialogs can

be inspected through a hierarchical process tree-

view, where the nodes provide links, opening the

exact task and email descriptions, including text and

attachments (Figure 2).

The collaborative functionality in CTM is further

supported through a notifications framework, which

issues notifications throughout the task delegation

hierarchies to inform participants in collaborative

processes if a related task of another process

participant is changed. Stakeholders can accordingly

adapt “in-situ” to the occurred changes.

3.3 Process Overview & Navigation

In CTM, process models emerge as examples for the

actual process execution and comprise the individual

to-do lists of all process participants. These lists are

integrated through the tracked task-related email

exchange. Thereby overall process models emerge

as Task Delegation Graphs (TDG) (Stoitsev et al.,

2008), where the personal task trees of different

users are shown in different user containers (Figure

3). We suggest that this overview provides a highly

intuitive process representation and enables end

users to more adequately recognize their position

and role in overall enterprise processes at a glance,

to identify potential bottlenecks and to evaluate

work distribution. Currently, due date, status and

percent complete indications are provided. The

description link within a task node opens a dialog

with full task (text) description. Tasks attachments,

added in OL tasks, are replicated in a central, DB-

based Artefacts Repository (AR) on the CTM server,

and are accessible in the task instances. Through the

“Show Roottasks” button the user can open a list

view with all initial process tasks (root tasks)

generated on the server throughout the whole

enterprise. Within this view the user can navigate

through the root tasks list and open a TDG (process

execution example) for a given root task.

3.4 Process Model Adaptation & Reuse

Within the presented paper a Task Pattern (TP) (Riss

et al., 2005; Stoitsev et al., 2008) is considered as a

reusable task structure, comprising one task with its

sub task hierarchy and the complete context

information of the contained task instances, like e.g.

description, used resources, involved persons etc.

CTM enables export of a local task from the

personal TDL to a single TP, and export of complete

TDG from the server to multiple TPs, which are

interlinked through suggestions according to the

delegation flow. A TP can be saved in a local or

remote Task Pattern Repository (TPR). A local TPR

is a XML-based document (Stoitsev et al., 2008),

whereas remote TPRs reside in a DB on the CTM

server. The exported task structures are managed in

the Task Patterns Explorer/ Editor (TPE) which is

shown in Figure 4. The TPE provides rich editing

and search functionality: cut, copy, paste, insert,

remove operations are enabled on task trees and on

Figure 3: Detailed process overview – Task Delegation Graph (TDG).

ENABLING END USERS TO PROACTIVELY TAILOR UNDERSPECIFIED, HUMAN-CENTRIC BUSINESS

PROCESSES - “Programming by Example” of Weakly-Structured Process Models

data in context fields (on the right hand side). TPE

enables also search and extraction of TPs from the

tracking repository. When editing the provided

process execution examples (interlinked TPs) in the

TPE “the user is not required to interact in the

interface domain of computational abstraction, but

works directly with the data that interests him or

her” (Lieberman et al., 2006). In this sense CTM

enables programming by direct manipulation of the

TP fields. The “Name”, ”Description” and

“Suggested Execution Time” fields hold simple task

context information in text format and are self-

explanatory. The “Owner” field recommends

expertise, i.e. when a TP is extracted from an

executed process, the owner is the person in whose

TDL a task was residing. The field “Suggested

Delegates” contains information about the persons,

who have the expertise to execute a given task.

When a TP is extracted from a collaborative process,

task recipients are set in this field. The “Suggested

Pattern” field holds a reference to a TP, which can

be used for the further processing of a task. In case

of TDG extraction, such references in requester

tasks point at recipient tasks, used for the further

task processing. The recipient tasks are themselves

extracted as separate TPs. Attachments to tasks are

represented as “Artefacts”. Custom adding of

artefacts to a task replicates these to the AR.

Studies on ad-hoc processes report that

“Employees often do not accept a strict sequencing

of those tasks which they have to execute

themselves, because this causes a limitation of their

flexibility” (Hermann, 2000). Our preliminary

studies confirm that statement and the necessity to

minimize sequencing of activities where possible.

Therefore we avoid the declaration of explicit

temporal relationships known from formal task

modelling approaches (Paterno et al., 1996; Veer et

al., 1996; John & Kieras, 1996) and formal

workflow modelling notations (OMG, 2006). TPs

provide structured process execution examples,

where the default assumption is to execute tasks

along the provided task hierarchy in a top-down

manner. Actual temporal relationships between tasks

can be observed only through the task statuses, e.g.

“Waiting for someone else”, “In Progress” provided

in the TDG in the web client during the concrete

process execution (see Figure 3). Automated

detection and export of temporal relationships to

design-time (TPE) is not provided currently,

although this may be useful for optimization of

rigidly recurring processes.

Figure 4: Task Pattern Explorer/Editor (TPE).

TPs can be reused through an “Apply Pattern”

operation, available on tasks in the TDL. It opens the

TPE, where the user can browse through different

TPRs and search for tasks on the server, based on

different criteria (owner, subject, description etc.).

Tasks from remote TPRs can be opened in the TPE,

whereas tasks from TDGs can be additionally

viewed in the web client so that users can estimate

the task applicability to their current situation. No

advanced proactive information delivery on tasks

(Holz et al., 2006) is currently provided. We have

considered that many users approach their

colleagues for help prior to looking for solution in

the available software infrastructure (see also Ribak

et al., 2002). Therefore TPs can be exchanged

through a “Send To” function in the TPE and as

attachments in task requests.

The application of a TP reactivates the process

example by generating the complete task hierarchy

and filling all pre-modelled structure and content

information in the TDL. If during execution a user

initiates a delegation, available delegates are

automatically suggested. A user can change the

anticipated (example) flow by entering different

recipients. Suggested TP references are also

available on tasks. A suggestion, stored as a

reference to a recipient task in the original process

execution, may be used by the person, activating the

TP, to accomplish the task themselves without

further delegations. If on the other hand a delegation

is issued, the recipient task contains the reference

ICEIS 2008 - International Conference on Enterprise Information Systems

and the recipient(s) can still refer to the suggested

TP to possibly adapt and reuse it. To allow this,

application of a TP from a local TPR enables

iterative replication of all referenced TPs from the

local TPR to a default remote, user-specific

repository, where these are accessible by all users.

3.5 Task Pattern Evolution

Best-practice deviations may occur due to changing

business conditions and different problem solving

strategies of end users. CTM provides functionality

to trace such deviations through task instance-based

ancestor/descendant relationships (Stoitsev et al.,

2008). Such are set e.g. on copy/paste of (sub) task

hierarchy in the TPE - iteratively each task in the

resulting hierarchy receives an ancestor reference to

the corresponding task in the original hierarchy.

When a TP is exported from an executed process

and saved to a remote TPR, all resulting tasks

receive ancestor references to the corresponding

original tasks in the tracking repository. If a remote

TP is applied, the resulting tracked tasks receive

ancestor references to the corresponding tasks of the

remote TP. If a TP is exported from an executed

process to a local TPR, the resulting tasks preserve

the information (id’s) of the tracked tasks. When a

local TP is applied, the resulting tasks receive

ancestor references to the originating tasks in the

tracking repository. Evolutions can be viewed in the

Task Pattern Evolution Explorer (TPEE) shown on

Figure 5. The “introduce consignment” task of user

Y (selected node) originates from a tracked ancestor

task with the same name, which was executed by

user X (root node). The latter task has also another

descendant, resulting from its reuse by user W (task

in the bottom). User Y has saved a global TP from

his execution to a remote TPR (expanded node with

black descendant icon under selected node), which

Figure 5: Task Evolution Explorer (TEE).

was reused in two further executions, the one of

which resulted in a second global TP version. The

TDG and dialogs of tracked ancestor/descendant

tasks can be shown through the “View in

Repository” button for case analysis.

4 CTM EVALUATION

The CTM evaluation was conducted at the textile

production company (cf. 2) and involved 6 users,

selected for having related, collaborative tasks:

Chief Officer Assistant (COA). serves as a single

point of contact to the chief officer (forwards

accept/reject) of contract proposals; coordinates all

departments (sales, IT etc.);

Chief Sales Officer (CSO). coordinates activities in

sales department, responsible e.g. for: internal

processing of special customer sales applications

(consignment, credits), credits approval, budget

planning;

Sales Employees (SL1 & SL2). process sales

orders, make credibility checks, participate in price

definition processes, assist CSO;

IT Department Lead (ITL). coordinates activities

of IT department, decides about acquisition of new

software and hardware; manages adaptations and

extensions to existing systems;

IT Employee (ITE). installs soft-and hardware;

executes business process-related transactions in

internal systems; maintains documentation about

executed transactions; provides guidelines for

transactions execution.

4.1 Setting & Extent of Use

The evaluation was initiated with a workshop in

which we gave a 1 hour presentation on CTM,

followed by 30 minutes individual training of each

user in the basic functionalities. Detailed CTM user

guides were provided to all participants. After

several days we visited the users individually to

check how they are working with the tool and to

provide further instructions. The evaluation

concluded with a short video recording and

transcription of the tool use, followed by a structured

debriefing interview, in which we asked each

participant to assess the basic features and to rate to

what extent CTM improved their ability to manage

tasks in ad-hoc processes using Likert scales and

freeform explanations.

The CTM trial was planned initially for 4 weeks.

However, the installation of the tool required

network adaptations as well as OL configuration

changes. Therefore only a 2 weeks trial was

possible. Problems with character encoding schemes

ENABLING END USERS TO PROACTIVELY TAILOR UNDERSPECIFIED, HUMAN-CENTRIC BUSINESS

PROCESSES - “Programming by Example” of Weakly-Structured Process Models

suspended the CTM usage by the COA for a further

week.

4.2 Findings

Despite the initial technical difficulties and usability

issues, mentioned in the following, end users found

the concepts behind CTM compelling and clearly

identified the high potential to structure and

optimize their activities with the tool - the average

overall approval rating for CTM was 4.29 (on a

Likert scale of 1: Hate it, to 5: Love it). A summary

of the observations follows:

Missing Initial Process Context. Some users

suggested that root tasks should be created by senior

employees, who actually trigger processes.

ITE: “I do not initiate processes, I actually execute

on them. […] I always expected to get a task request

from somebody [COA, CSO] who would create a

root task and distribute the sub tasks. I then would

receive a task, break it down and distribute the

resulting tasks to the others [Sales].”

Due to the encoding problems in the TDL of the

COA, the latter did not send requests for a week

after ITE had started using CTM. This affected also

the amount of tasks ITE acted on. Similarly, SL1

had created a root task for a task description, which

was sent by CSO per email some time ago but was

not acted upon before the CTM installation. No root

tasks were created for ongoing activities in which

users were engaged before CTM installation. This

reveals that process modelling can be triggered

along the organizational hierarchy, where senior

employees can drive a top-down implementation of

the “Process of Me” (Gartner, 2006).

Transparency. The ability to represent artefacts in

process steps was considered crucial. We

encountered that different artefact versions were

attached to consequent tasks in a process flow,

which revealed how artefacts are elaborated within a

process. For example an empty, preformatted MS

Excel table was attached in a request issued from

CSO to SL2, and a filled MS Excel table was

available in the resulting SL2 recipient task, which

was elaborated to 75%. Further, users highly

approved status information and notifications on

task changes as they saw in them the potential to

reduce overload for calling colleagues and writing

emails with task status enquires.

SL1: “Such processes [price definition] draw like a

red thread through the whole company. I certainly

want to know how far things have gone. […] It is

annoying when you do not get feedback on

requested actions. This [CTM process overview]

will save me the effort to constantly call people or

write mails to ask about the status of things.”

Generally, employees with managerial functions

had greater interest in the overview functionality

than others. SL2 for example surprisingly stated that

seeing what others do might not be of interest to him

as it might concern activities outside of his expertise

scope. COA, CSO, SL1 (who had more senior

functions) and ITL clearly wanted an overview.

As CTM was used only by a small group of

people, privacy issues were not raised during the

trial. However ITL stated that authorization has to be

considered for extended CTM use in the enterprise

by providing the possibility to hide certain process

fragments in black-box containers in the web

process overview. SL1 further demanded extensions

in the notifications handling and suggested e.g.

having notifications on each change in a delegated

task and its sub tasks – structural or context change.

Notifications for overdue of delegated tasks were

also requested. As a further extension, users

suggested summing up percent complete of sub tasks

and increasing the percentage of a parent task.

Structured Storage and Retrieval of Process

Knowledge. Users generally reported that creating a

task in the TDL does not impede their current work

practice compared e.g. to dealing with email.

SL2: “A task is a task - I clearly know that I should

act on it. […] Putting it in the CTM task list does not

bother me. I need to think how it should be handled

anyway. If I can explicitly write that down, this only

helps me to clearly structure my thoughts before

executing and reduces the chance to miss

something.”

ITE further reported, that sometimes CSO asks

him to execute transactions, which he is normally

not allowed to. Before the CTM installation, ITL

would preserve the emails, requesting those

transactions, for responsibility tracking. Receiving a

CTM task for such transactions reflected this

“opportunistic” behaviour in the generated process

example (TDG) on the server and hence in the

emerging process model. Despite the clear benefits

from CTM usage for visibility on time-critical

activities, users stated that email cannot be replaced

fully by CTM tasks. Informal enquiries outside of a

concrete process would still be done over email.

Although only several TP were extracted – 2 in

IT department (1 in a remote TPR and 1 in a local

TPR) and 3 in sales (1 in a remote TPR and 2 in

local TPR), the benefit from structuring process

knowledge in a way that it could be reused was

stated as a clear benefit. However, we clearly

perceived that users were uncertain about the reuse

potential of TP and the way these should be

ICEIS 2008 - International Conference on Enterprise Information Systems

distributed to others. The overall attitude was that

global TP should be delivered by a (senior) domain

expert, who can handle also the responsibility for

providing them. CSO e.g. experimented and

developed a TP on a remote TPR instead of writing

a text-based guideline. SL2 on the other hand

refrained from submitting a TP on a remote TPR

while stating that he could send the local TP to a

colleague personally upon request and furthermore,

that he “silently agrees” for other colleagues to take

and adapt his implicitly generated task example from

the tracking repository on their own responsibility.

Some of the users proposed that the collaborative

flow on tasks should be structured better to facilitate

the handling of CTM emails for task delegation. The

“Move CTM’s” functionality (cf. 3.2) was not

accepted well - users preferred to get CTM request

messages in a dedicated “CTM Mail/Requests”

email folder and responses in a “Responses” folder.

Exchange of Process Knowledge. Having an

example of how a problem should be approached

was appreciated by all users.

SL2: “Basically I have to achieve certain output for

the tasks I receive [from CSO]. I really appreciate to

know how she would break down the task and what

the different facets in the task are. This helps me to

stay on the right track and to know what is expected

of me.”

However, we actually observed that CSO would

send a single task with generic description e.g.

“prepare contracts for customers C1, C2, and C3”

and SL2 would then break it down, creating a task

for each customer. Therewith tasks disperse and

refine by falling through the organizational

hierarchy. This reveals that “seeding, evolutionary

growth, and reseeding (SER)” (Fisher et al., 2004)

towards complementing abstract process

descriptions can happen during task execution and

iterative reuse of process examples in organizations.

Domain experts, e.g. ITL, on the other hand did

not think that they would benefit much from external

knowledge. ITL however appreciated being able to

distribute knowledge himself i.e. as TP on a remote

TPR, to avoid repeated inquiries from other

employees on same topics.

Connecting Best-practices and Running

Processes. The users considered that comparison of

TP and running tasks, resulting from their

application, might not scale for large processes.

Best-practices were generally desired as higher-level

process descriptions, while running processes could

produce multiple fine-grained tasks.

CSO: “As far as I am concerned a TP will contain

only top-level tasks as my employees always do

things differently. This doesn’t bother me if the

results are delivered on time. […] It is good to have

a guideline, even if you do not care how the

described tasks are accomplished concretely.”

The overview provided in the TEE was not

considered intuitive. Differences in task structures

could be identified through additional effort, which

would bring benefit only to managerial employees.

Users suggested enabling task comparison in

“swimming lane” overview, where the

corresponding top-level tasks can be put against

each other. This would enable users to better see the

corresponding and missing process facets, by

possibly discarding the low level tasks. For the

latter, filtering techniques based on different criteria

like e.g. “Task Level” and “Owner” were suggested.

Tracing of Evolving Best-practices. Despite of the

deficiencies in the TEE usability, the functionality it

provided was considered necessary by senior

employees due to the frequent changes in informal

process recommendations. Tracing of such changes

could help to at least undo wrong strategies.

SL1: “We often change processes to check if we can

achieve better results. We check e.g. for the

processing of these contracts we needed that much

time, while we have planned that much. […] If we

see that a change does not deliver better results, we

switch back to our previous practice. […] An

overview and comparison of the tasks for both

practices in CTM is nice to have.”

With this respect the provided structural

overview was still insufficient as users cared also

about certain performance indicators.

Users proposed that the comparison of task

hierarchies in TEE should be enabled based on

specific criteria like e.g. execution time, persons

involved. It was further suggested that in addition to

the ancestor/descendant relationships also versioning

of TP should be supported.

5 CONCLUSIONS & FUTURE

WORK

The presented paper describes an integrated

approach for leveraging user experience with email

and to-do lists and ensuring a “gentle slope of

complexity” for process tailoring by end users. It

delivers a valuable extension to known evolutionary

workflow approaches by enabling “programming by

example” of decentralized-emerging, weakly-

structured process models by both: users - executing

processes, and domain experts - explicitly adapting

captured process examples. Thereby SER of weakly-

structured process models is enabled through the

top-down implementation of the “Process of Me”,

ENABLING END USERS TO PROACTIVELY TAILOR UNDERSPECIFIED, HUMAN-CENTRIC BUSINESS

PROCESSES - “Programming by Example” of Weakly-Structured Process Models

where: (i) generic tasks refine during execution; (ii)

users can adapt reusable process fragments (TPs)

through direct manipulation of the execution data

(delegations, artefacts, suggested TPs). Thereby

opportunistic and emergent changes are supported

during runtime and design time. CTM captures

conversational (email) and control (task) flows.

Unlike known email-based workflows, CTM

provides the ability to decouple process fragments

(interlinked TP) with different granularity from

process runtime representations and to make them

available for SER by managing task instance-based

ancestor/descendant relationships, allowing

navigation to the original or to similar execution

contexts and inspection of task-related dialog flows.

The CTM evaluation delivered user-proposed

extensions which will be addressed in further

prototype implementations. Long term evaluation in

the partner companies is under negotiation and will

allow the generation of larger tracking and TP

repositories and their quantitative evaluation as well

as scalability assessments. Further research will aim

at the translation of user-defined process fragments

to known formal process modelling notations

towards automation of rigidly recurring processes.

ACKNOWLEDGEMENTS

The work, this paper is based on, was supported

financially by the German Federal Ministry of

Education and Research (project EUDISMES,

number 01 IS E03 C). We thank to all participants in

the user studies for their time and cooperation.

REFERENCES

Agostini, A., De Michelis, G., 1997. Rethinking CSCW

systems: the architecture of Milano. In ECSCW’97,

Springer, pp. 33-48.

Bellotti, V., Dalal, B., Good, N., Flynn, P., Bobrow, D. G.,

Ducheneaut, N., 2004. What a To-Do: Studies of Task

Management towards the Design of a Personal Task

List Manager. In CHI’04, ACM Press, pp. 735-742.

Bellotti, V., Ducheneaut, N., Howard, M., Smith, I.,

Grinter, R., 2005. Quality Versus Quantity: E-Mail-

Centric Task Management and Its Relation With

Overload. Human-Computer Interaction, Lawrence

Erlbaum Associates, Vol. 20 (2005), pp. 89-138.

Button, G., 1994. What’s Wrong With Speech-Act

Theory. Computer Supported Cooperative Work, Vol.

3, No.1 (Mar. 1994), Springer, pp. 39-42.

Fischer, G., Giaccardi, E., Ye, Y., Sutcliffe, A.,

Mehanjiev, N., 2004. Meta-Design: A Manifesto for

End-User Development. Communication of the ACM,

Vol. 47, No. 9 (Sep. 2004).

Gartner Research, 2006. Person-to-Process Interaction

Emerges as the 'Process of Me’. Gartner Inc.

Herrmann, T., 2000. Evolving Workflows by User-driven

Coordination. In D-CSCW’00, Teubner, pp. 103-114.

Holz, H., Rostanin, O., Dengel, A., Suzuki, T., Maeda, K.,

Kanasaki, K, 2006. Task-based process know-how

reuse and proactive information delivery in

TaskNavigator. In CIKM’06, ACM Press, pp. 522-

531.

Jorgensen, H. D., 2004. Interactive Process Models. Ph.D.

Thesis, Norwegian University of Science and

Technology, Trondheim, Norway.

John, B., Kieras, D., 1996. The GOMS family of analysis

techniques: Comparison and contrast. ACM

Transactions on Computer-Human Interaction, Vol. 3,

No. 4 (1996), pp. 320-351.

Lieberman, H., 2001. Your Wish is My Command:

Programming by Example. Morgan Kaufmann.

Lieberman, H., Paterno, F., Wulf, V., 2006. End-User

Development, Springer.

MacLean, A., Carter, K., Lövstrand, L., Moran, T, 1990.

User-tailorable systems: pressing the issues with

buttons. In CHI’90, ACM Press, pp. 175-182.

Object Management Group, 2006. BPMN,

http://www.bpmn.org/

Paterno, F., Mancini, S., Meniconi, S., 1997.

ConcurTaskTree: a diagrammatic notation for

specifying Task Models. In Interact’97,

Chapmann&Hall, pp. 362-369.

Ribak, A., Jacovi, M., Soroka, V., 2002. “Ask Before You

Search” Peer Support and Community Building with

ReachOut. In CSCW’02, ACM Press, pp. 126-135.

Riss, U., Rickayzen, A., Maus, H., van der Aalst, W.M.P.,

2005. Challenges for Business Process and Task

Management. Journal of Universal Knowledge

Management, Vol. 0, No. 2 (Nov. 2005), pp. 77-100.

Schwarz, S., Abecker, A., Maus, H., Sintek, M., 2001.

Anforderungen an die Workflow-Unterstützung für

wissensintensive Geschäftsprozesse. In

WM’01, 1st

Conference for Professional Knowledge Management

Baden-Baden, Germany.

Stoitsev, T., Scheidl, S., Spahn, M., 2008. A Framework

for Light-Weight Composition and Management of

Ad-Hoc Business Processes. LNCS 4849, Springer.

Veer, G. v. d., Lenting, B., Bergevoet, B., 1996. GTA:

Groupware task analysis - modeling complexity. Acta

Psychologica, pp. 297-322.

Wiig, K. M., 2004. People-focused knowledge

management: how effective decision making leads to

corporate success. Elsevier Butterworth–Heinemann.

Wulf, V., Jarke, M, 2004. The Economics of End-User

Development. Communications of the ACM, Vol. 47,

No. 9 (Sep. 2004).

ICEIS 2008 - International Conference on Enterprise Information Systems