Supporting document management in complex multitask environments

Supporting document management in complex multitask environments

Olha Bondarenko (TU Eindhoven)Ruud Janssen (Novay)

Samuël Driessen (Océ Technologies)

Goodbye to paper!

• problems with paper– costs of storage– static content, hard to revise– no remote access– large ecological footprint– image: “old fashioned”, “chaos”

• since 1990’s a steady move to digital work styles– distribution via email– archiving in doc. mgt. systems– flexible offices, clean desk policy– mobile devices, nomadic workers– connected anytime, anywhere

Or is paper is here to stay?!

• paper documents abound in many modern offices

why is paper still around?

• paperless office has been “an inch away” for decades

what does paper support that digital technology still doesn’t?

“The Myth of the Paperless Office”

• Sellen & Harper, Xerox PARC, 2001

• paper supports knowledge work– authoring, reviewing– planning and thinking– collaborative activities– organizational communication

• paper has unique affordances for reading– flexible navigation, rearranging to create an

overview– annotation, interweaving reading and writing

Our own research

• started in 2002 at Océ, continued at TU/e• originally application oriented

– new ways to integrate paper and digital technology

– learn from paper to improve digital technology

• research divided into three stages1. how does paper support task management?2. how does paper support task switching?3. how does paper represent context and task

state?

Three challenges

• Information overload & fragmentation– Increased number of communication channels &

storage– Negatively affects performance and causes stress

• Multitasking & frequent switching– Information workers have an unstructured workflow– Working on multiple tasks and switch before

completion

• Mobility demand– Flexible offices and nomadic work– Digital support is needed independent of physical

location

Stage 1

how does paper supporttask management?

Motivation and approach

• paper seems to support knowledge workers in their task management, but in what ways?

• contextual interviews, 28 participants– semi-structured interviews, in situ– artifact walkthrough– critical incident collection– taking pictures, making sketches

• data analysis– interpretation using affinity diagrams– requirements for document management systems

Findings at a glance

• a wide variation– desks vary from clean and well ordered to one big

mess– workers apply a mix of “filing” and “piling”– workers vary from “paper traditionalists” to

“digital frontrunners”

• but also many similarities– paper documents: hot and warm versus cold– paper piles support task management in many

ways– knowledge workers need to manage a mix of

paper and digital documents

hot documents

“I am working on this now”

warm documents“I need to work on

this soon”

cold documents“this is old stuff, I may

still need it in the future”

hot documents“I have to finish

this today”

warm documents“this is lying around

from yesterday”

The person behind the desk

• knowledge workers are multi-taskers

• they need to switch between tasks

• they need to postpone tasks

• they need to be reminded about tasks

• they need an overview of their tasks

• document management is not their primary concern

Paper piles support task management

• piles provide an overview over tasks

• a pile serves as a reminder for a task

• a pile keeps individual documents in their task context

• a pile can be rearranged to reflect changes in a task

• a pile can be relocated to reflect changes in a planning

• piling requires no verbalization effort

The effects of ongoing digitization

• from mail to email• from desk to

desktop• multiple incomplete

systems

• paper and digital still coexist

• printing is easier than scanning

• printing to combine documents from various sources

Least effort principle

• Document management is a secondary task for most of the information workers

• A document management system has to require as little effort as possible

• Information workers are already overloaded with places where documents need to be stored and maintained

• A document management system has to be based on a system that is already in use or it will increase fragmentation and fail

Grouping documents

• Multiple places to store documents contribute to information fragmentation and increase overload

• Documents from different sources need to be brought together when working on a task

• Paper is still the only tool to support this

Requirements for technology

systems to support document management should1. place documents in their task related context2. combine documents from various sources3. provide an easy way of (re)grouping

documents4. allow flexible and ad hoc restructuring within

a group of documents5. be based on a system that is already in use6. require as little cognitive effort as possible

Stage 2

how does paper supporttask switching?

Motivation and approach

• piles and documents contain cues that reflect the state of interrupted or postponed tasks

• how are these cues created or changed?

• in situ observation, 4 participants– observer unobtrusively present at workplace– recording all actions and changes in the environment– pre- and post observation interviews

• data analysis– identify tasks and task switches– identify task switching strategies that aim to preserve

the state of the interrupted task– identify factors influencing the task switching strategies

Findings at a glance

• 8 task switching strategies were identified– active or passive– cues or no cues left in the environment

active strategies

passive strategies

cues left in environment

• activation• changing location• gathering• writing down• printing

• no action/state preserved

no cues left in environment

• deactivation • no action

Task cues left in the environment

• t1 represents the current task (no action/state preserved)• t2 are documents she collected for an upcoming meeting (gathering)• t3 are documents that she put aside to work on later (changing

location)

Task switching: factors

Reason for switch Domain of the last action

0 10 20 30 40 50 60 70 80 90 100

None of the mentioned

Obs. disagreement

No action

No action/ state preserved

Printing

Writing down

Gathering

Changing location

Activation

Deactivation

%

EXTERNAL (N=35)

INTERNAL (N=148)

0 10 20 30 40 50 60 70 80 90 100

None of the mentioned

Obs. disagreement

No action

No action/ state preserved

Printing

Writing down

Gathering

Changing location

Activation

Deactivation

%

EXTERNAL (N=35)

INTERNAL (N=148)

0 5 10 15 20 25 30 35 40

None of the mentioned

Obs. disagreement

No action

No action/ state preserved

Printing

Writing down

Gathering

Changing location

Activation

Deactivation

%

DIGITAL (N=90)

PHYSICAL (N=79)

0 5 10 15 20 25 30 35 40

None of the mentioned

Obs. disagreement

No action

No action/ state preserved

Printing

Writing down

Gathering

Changing location

Activation

Deactivation

%

DIGITAL (N=90)

PHYSICAL (N=79)

Internal vs external reason to switch

• external reasons: passive strategies dominate– passive strategies allow for an immediate switch– cues serve as a reminder but are fragile– supports task resumption when task is resumed on

short notice and environment remains untouched

• internal reasons: active strategies dominate– active strategies involve creation of more

persistent cues (location, structure, explicitness)– more robust cues that will survive longer, at the

cost of some time and effort

Paper vs digital switching

• Task switching in the physical world– Active patterns dominate– People actively rearrange their documents at

the moments of task switching

• Task switching in the digital world– Passive patterns dominate– People switch from one task to another without

taking an observable action to preserve the state of the task

Multitasking & switching

• Task-related document collections– Combining documents from different sources together– Easily (re)groping documents as the task goes

• Stable state– Regrouping is a mechanism to indicate the stable

state within a task– Stable states play a major role in successful task

resumption– Gives an answer to the question “where was I” in a

task?– In paper pile, the most recent document is usually on

top, regrouping and rearranging documents within groups is very easy

Stage 3

how does paper representcontext and task state?

Motivation and approach

• piles and documents contain cues that reflect the state of interrupted or postponed tasks

• what are these cues, and what do they encode?

• triad elicitation, 12 participants– pictures of participant’s desk and objects on it– randomly select sets of 3 pictures– “choose two that are similar, and different from

the third”– laddering down (“what makes you think that…”)– laddering up (“what does it mean for you that…”)

Approach (continued)

data analysis• visual cue: statements describing phenomena

that can be observed by other people• semantic judgment: statements describing

judgments based on knowledge of the participant

1. statements were classified2. statements within a class were categorized3. visual cue categories were clustered4. semantic judgment categories were clustered

• “these two are printed” “that one is handwritten”[initial statement; visual cue: text presentation]

• “the text is regular and evenly spaced” “the text is irregular”[laddered down; visual cue: shape of content]

• “these I will give to others” “this one is to remind myself”[laddered up; semantic judgment: interaction purpose]

Example: triad elicitation and laddering

Triad elicitation study results

• Recognized visual cues– Identifying what people see when they look at

their documents

• Encoded semantic judgments– Identifying what people think when they look

at their documents

• Model of relationships– Bringing seeing and thinking together

Findings at a glancecolor contrast

structureof layout

appearanceof cover

textpresentation

contentpresentation

visibility ofcontent

shape ofcontent

size anddimensions

volume

grouping

position inthe group

structure ofthe group

edges

relativelocation

absolutelocation

relation toother objects

ownership &confidentiality

interactionpurpose

informationpresentation

keepingdecision

throw-awaydecision

similarityand relations

periodof use

resourcedemand

actiondemand

currentusage

frequencyof access

importance

what does it mean for you

what makes you think that

what does it mean for you

what makes you think that

what does it mean for you

what makes you think that

location

group arrangement

page content

purpose

task context

planning & overview

Getting grip on the relationships

• page content:– extent to which cues can be changed without interfering

with work itself, is limited– cues are explicit and changes require a lot of effort

(such as verbalization)

• group arrangement:– cues range from explicit (plastic folders, paper clips) to

implicit (loosely gathered piles)– explicit cues are more persistent and require more effort

to change

• location:– most flexible cues when it comes to assigning a

meaning– a natural reference point (the self, other objects)– easiest to manipulate, it happens almost automatically

Encoding semantic information

• Tasks are complex notions in the head and parts need to be offloaded in the environment

• Information about the state of the task is encoded by associating available visual cues with semantic judgments about the task

• This is one of the major reasons for paper being on the desks

Easily available and flexible cues

• The mechanism of encoding information about the task state needs to be transferred to the digital domain

• Digital environment is very different from physical and the mechanism cannot be copied directly.

• Encoding information in the digital domain often requires significant effort compared to the physical domain

• The visual cue should be comparably salient and easy to adjust

Requirements

Overview

Task decomposition level

Supported process

Requirements

Knowledge work Document and task management

Support least effort principle: document and task management is a secondary task for an information worker

Unstructured workflow

Non deterministic task flow

Support unstructured workflow with ad-hoc collections of associated documents: documents are representation of tasks

Multitasking Frequent switching

Support representing the current state of the task in the environment: tasks are most often switched without finishing up

Task suspension / resumption

Stable state creation

Support encoding of semantic information about the task state into easily available and flexible visual cues

Supporting knowledge work

• Least effort principle dominates– Based on the system in use– Support variety of styles and needs

• Paper– Supports variety and unifies document sources– Increases fragmentation and does not support

refinding

Supporting unstructured workflow

• Provide ad-hoc collections of task-related documents as representations of tasks– Arrange documents along with the task flow– Provide overview within and across collections– Allow implicit planning / prioritizing

• Paper– Provides a constant overview of tasks in a

glance– Provides unobtrusive reminders– Quickly gets chaotic and outdated

Supporting multitasking

• Support task suspension and resumption in the environment– Preserve and reflect the current task state– Allow custom changes in the representation

• Paper– Allows for robust task state representation– Provides an overview of a number of tasks – Always requires action

Support task suspension / resumption

• Support encoding of semantic judgments in digitally available environmental cues– Extract information about the past (automated)– Provide information on a proper level– Allow user input to encode information about the

future– Encode information on a proper level / effort

• Paper– Supports easy & flexible encoding and decoding

of cues– Cues are highly implicit and prevent long term

planning

Summary and conclusion

learning from paper toimprove digital technology?

Lessons learned

• Design according to the fundamental user needs– Technologically driven design complicates

interaction and increases user effort – Superficial requirements only suitable for later

design stages

• Take into account human capabilities and constraints– Find the ways to use advantages of the digital

domain rather than mimicking physical cues

Future research

• Investigate the existing encoding mechanism in the digital domain

• Investigate social, organizational and personality aspects

• Iterative design to evaluate and refine the requirements