Title: Supporting Administration in Asynchronous Team Environments
1Supporting Administration in Asynchronous Team
Environments
Roger Tagg School of Computer and Information
Science University of South Australia
2Agenda
- Introduction
- University Teaching Management case study
- Workflow approach
- Learning support approach
- Knowledge management approach
- Data based approach
- User interface approach
- Software Agent approach
- The future of groupware
- Concluding remarks
3Introduction
- IT has yet to bring a major improvement in
general administration (we are still all snowed
under!) - We do have many domain-specific applications and
task-specific tools - The limiting factor now is ability to integrate
and cross-relate them all - Group support is still weak
4Case Study Academic Teaching Management
Also involved Administrative staff (School,
Division, Uni), Teaching Quality controllers,
Exam organisers etc.
5Characteristics of our Case Study
- Meeting the calendar deadlines is first priority
- Following a fixed process is not always important
(not written down anyway) - Processes vary for each campus
- Campuses have varying degrees of autonomy
- Many staff come and go (and need help)
6My own experience with our Case Study
- I joined and immediately had to take on 2 courses
I hadnt seen before - I had no knowledge of CIS procedures and no-one
had anything I could read - Emails hit me demanding this and that
- I had an Assistant Lecturer who had helped once
previously - Few colleagues had any time to answer my
questions - I wasted a lot of time
7Some of the Procedure Manuals
8Workflow Approach
- I have been a workflow researcher, so maybe I
should take a dose of my own medicine - We started before we had any tools
- (now have Adonis (BOC, U Vienna) and
FlowMake/Chameleon (DSTC, U Queensland)) - We charted the processes as we interpreted them
from manuals and tacit knowledge
9Process Model Run an Exam
10Process Model Assignment
11Process Model Course Offering
12Process Model Run a Course
13Flexible Workflow
- A teams (or individuals) use of Workflow can be
looked at in 3 quite different ways - As a driver of a mission-critical business
process that must be performed in correct
sequence ? Big workflow - As an integrator of software components in an
customer-assembled mega-system ? Enabling
workflow - As a set of rules applied to defined events in a
cooperative system ? Small workflow
14Problems with Workflow as a solution for our case
study
- Processes are often too volatile
- A team often starts without any!
- There are too many exceptions
- Most team members dont have skill in modelling
processes, so cant apply dynamic changes - It is difficult to enforce processes beyond one
organizations boundaries (associated Unis run
the overseas campuses)
15Can workflow be optional?
- A possible spectrum is
- Fully prompted all work is done in response to
a worklist item - Opt out user decides not to follow the prompts
- Opt in user chooses to be prompted by the Wf
engine - Email prompted, but no compulsion
- Fully ad hoc user totally decides what work to
do next - Problem is, mix and match may not be an option
- Next performer cant rely on the process state
having being updated - Cant monitor, audit or analyse performance
16Comments on the Workflow Approach in our Case
Study
- The modelling conventions and tools do not easily
support multiple parallel tasks of the same type
(ref van der Aalst) - Few sequence restrictions were really mandatory
- The process would always be overridden to meet
deadlines - No staff showed much enthusiasm in having their
work allocated by a software system - There are some team members who dont participate
in any information sharing- especially if any
software comes from Microsoft!
17Learning Support Approach
- The problem can be interpreted as providing
learning support for inexperienced team members - In our case, course coordinators may stay for
years, but most others only do the job for 1 or 2
years - The process perspective cannot be ignored though
18What are the main problemsfor Inexperienced
Teams?
- Not knowing the procedure for doing something (or
if one exists) - Not knowing who to ask
- Cant find the relevant data
- Different tools and applications that dont
integrate - Team members who dont participate in the
information sharing (again!)
19Some Paradigms for Supporting Inexperienced Teams
- Process Help
- Guided process execution (hand holding)
- Tacit knowledge exchange facilitation
- Knowledge Management
- Intelligent Agents
20Process Help Possibilities
- Answering the question how the XXXX do I do
this? - Hypertext-encoded procedure manuals (XML)
- Big backlog of existing manuals to re-encode
- Diagrams with you are here arrows
- E.g. searchable activity descriptions and
superimposed arrows - Reference to a file of who to ask experts (part
of Tacit Knowledge)
XXXX
21Process Help
22Process Help Diagram Example
You are here
23Guided Process Execution
- Answering the request take me through this step
by step - Is less than general application of a workflow
process template - Single process instance only, from one
performers viewpoint - Needs comments in addition to the activity
description - Extracts from the full procedure manual
- Guidelines on how to do it, quality expected,
related policies - Needs to include the activities this performer
ISNT doing as well as those he/she IS doing - Ideal would be a variable Wizard, whose steps
were generated from the process template - Context (primary key) should ideally be deduced
from the users previous actions
24Tacit Knowledge Exchange
- Essentially, a team bulletin board plus an
inventory of who knows what - FAQs
- Index and search facility for data in web pages
and files - Working documents building up process knowledge
as it solidifies
25LiveNet (U Tech Sydney)
- LiveNet has
- About this workspace (a user can join many)
- GoalsNewsSurprisesMilestonesTerminologyFAQ
- About your role
- Folders Chat Room(n)Â
- Communication Send email
- Who to ask list
26LiveNet Screenshot of a Light Workflow Rule
Wizard
27LiveNet - Observations
- Specifically supports team working
- Aims to support Tacit Knowledge Exchange
- Includes some small workflow
- Includes some Process Knowledge Management
- Tested in a cooperative business proposal
preparation and review example (Ingrid Slembek) - Agents are subject of current research (Alan Lin)
- --------------------------------------------------
---------------------------- - Doesnt seamlessly integrate with ordinary email
- Doesnt have a component customization system for
integrating with external applications - Isnt dashboard-friendly or mobile-friendly
28The Prattle Syndrome
- To prepare for a planning retreat, I set up a
shared workspace (bulletin board, discussion
forum etc) - Team members were too pressed for time to do more
than answer their email, phone and F2F
emergencies - So they rarely or never got to log on to the
discussion forum - Pull was inappropriate for over-busy teams
- We needed a Push approach
- Alerters, Sticker
- Generate urgent email messages
- Post tasks to a unified Task List
29Knowledge ManagementApproach sub-project in
LiveNet
- A team may start with few, or only very broad,
established processes - If actual activities are recorded, then do
- Process Mining detecting possible patterns in
events - Learning interacting with team members to
establish patterns as approved processes - Ref Biuk-Aghai, Simoff Slembek,
Knowledge-Assisted Reverse Engineering of
Virtual Work Processes
30Data Based Approach
- If the administrative process is asynchronous
enough, one can take the view that information is
exchanged by simply storing it in a shared
repository - One may still need to add alerters
- Document other shared data management may be
more important than processes - In some cases, lack of established trust may
enforce Send rather than Share
31The Data Perspective
- Data may exist in many topic types, e.g.
- Status of the entities being worked on
- Decision support data, statistics etc
- Metadata what types of data are stored?
- Negotiating positions
- Data exists in a lot of different software forms
- Application databases
- Office tool file types
- Web pages and XML
- A lot of the data is still unstructured and
difficult to link to the more structured data - Terminology is an issue ( hence Ontologies)
32Universal data filing
- All types of data should be filed by what they
relate to, not by what software they were created
in (current Groupware folders problem) - All types of data should be relatable by their
content (e.g. emails with database rows with word
documents with web pages) - If a piece of data is relevant to multiple
categories, it should be accessible (either by
replica or alias) in all those categories
33User Interface Approach
- However good the rest of a system is, if the user
interface is no good no-one will use it - Users may be working on many types of task
- No good having too many different interfaces
- The Outlook syndrome
34Integrated Computer Workspaces
- Dashboards
- UI containers that assemble a users total work
- Unified UI objects
- e.g. Single task list (inbox workflow tasks
) - Integration between standard team support
components work management - Mainstream groupware (email calendar tasks),
workflow enabling, project management, contract
management, time tracking, common office tools - Integration with domain-specific applications
- e.g. accounts receivable, production planning, HR
35Dashboard Example
36Dashboard - Implications
- Need to provide a tailoring facility for creating
and pasting in compact controls - If a control gets added to - or removed - from a
dashboard, the system should manage things - Need to capture knowledge about the tool(s)
underlying that control - Need to distinguish shared and individual
elements on the dashboard - Screen space may be a limitation
- Multiple displays
- Mobile-friendly versions?
37Drop Boxes
- Are dashboard icons (or sub controls) that act
as specialized administrative assistants - Are supported by rule based components (agents?)
that work out what to do when something is
dropped - Invoke a categorization agent to deduce, (e.g.
the parameters than determine the rules governing
action to be taken, including where to file
things) - Support many types of droppable things, e.g.
whole messages, index lines, selected text /
graphs, file icons (including attachments)
38Drop Box for Incoming Mail
39Software Agent Approach
- A degree of intelligence is often necessary to
enable software to deal automatically with things
dropped into Drop Boxes - User Agent (one that learns about its master
i.e. beyond the MS Office paperclip) - Categorization Agent
- Puts any object into categories based on its
content, and possibly the current context - Filing Assistant or Agent
- Suggests a place to file everything based on
content, team goals, individual goals - Are these all genuinely intelligent Agents, or
are they just Components or Services?
40User Interface Agents
- Learn user preferences
- Learn about the users categories, priorities,
most likely actions - for the individual
- for workgroups he/she belongs to
- Exchange knowledge with other agents / users
- Complete delegated tasks autonomously
- Automate tasks at a private level
41Categorization Agents
- Have been used successfully to classify
documents, including email and spam in
particular - Typically, mix some categories provided by the
user with what the user interface agent learns - There are a variety of categorization algorithms
no universal best approach yet - But
- no real plug and play tool to help file documents
in the categories
42Multiple Drop Boxes Supported by Common Agents
43The Future of Groupware
Current Groupware (e.g. Exchange/Outlook, Lotus
Notes), in spite of many research efforts, have
stagnated for a number of years
- E-mail messages are not effectively linked to
calendar appointments tasks - Outlook is not easily linked to application
systems databases - Outlook/Exchange folder storage system is quirky
and may not fit the organizations backup policy - Too much to learn users use only 10
- Too much work to configure rules
- Not everyone is willing to use what Microsoft (or
Lotus) offers
44Groupware Market Mismatch (adapted from Sheth)
Process-free Groupware
BigWorkflow
45Why we ought to expect a New Generation of
Groupware
- A lot of administrative work falls into the
middle of the last graph - There has been no real dent so far in information
overload - Depth of use of Groupware tools is very low not
much apart from email - Virus and spam attacks have been a big problem
- Clients have inconsistent features (3 types of MS
Outlook) - Users arent managing to integrate Groupware data
with application data - Vendors (and others) have been researching into
agents and categorization for some years
46Concluding Remarks Autonomy of Team members
- One organization can rarely control how team
members in other organizations do things - In some University CS departments, even internal
control doesnt stand much chance! - Some individuals may not participate, even in a
prescribed common environment - The dynamics of power in teams may be more
important than the technology Riempp 1998
47Is Government Administration Something Special?
- Correctness and fairness may be more important
than cost and speed - The individual users performance assessment may
loom very large - Knowledge is power and sharing it dilutes a
managers power base - Is there a culture of knowledge sharing?