How Do People Get Back to Information on the Web? How Can They Do It Better? William Jones, Harry Bruce The Information School University of Washington.

How Do People Get Back to Information on the Web? How Can They Do It Better?

William Jones, Harry BruceThe Information SchoolUniversity of Washington

Susan DumaisMicrosoft Research

The Problem Finding things is a well-studied

problem. Keeping things found is not so well-

studied but arises in many domains: Everyday objects in our lives Personal files – paper and electronic Email The Web

Related Work Organizing personal files

Files & “piles”, (Malone, 1983) Location memory is limited, (Jones &

Dumais, 1986) Preference for browsing, (Barreau & Nardi,

1995); but see Fertig, Freeman & Gelernter (1996) for a rebuttal.

Organizing email Similar use patterns, similar problems as

for personal files, (Whittaker & Sidner, 1996)

Related Work (cont.) Organizing the Web

Widespread use of “Bookmarks”, (Pitkow & Kehoe, 1996)

Steady increase in number with time, (Abrams Baecker & Chignell, 1998)

Increasing use of folders, (Abrams et al., 1998)

Frequent use of “Back” button within a session; infrequent use of “History”, (Tauscher & Greenberg, 1997)

Overall Research ObjectivesWe’re looking for answers to the following questions:

How do people manage information for re-access and re-use? How do people “keep found things found”?

What problems do people encounter?

What can be done to help?

The Research Project Study 1: Observe “keeping” activities as

participants complete work-related, web-intensive tasks in their workplace.

25 participants in all.

Study 2: Observe efforts to “re-find” web information for a subset of these same participants.

13 participants in all

Survey a larger group. Initiated.

Prototype selectively.

Study 1: The Participants 6 Researchers. 10 Information professionals -- including

librarians and corporate information specialists.

9 Managers.

Study 1: The Procedure Prior to the observation

Participants completed an email questionnaire… and listed at least three work-related, web-intensive “free-

time” tasks. One task was selected for the observation.

During the observation Participants were observed in their own workplace. Sessions lasted about an hour. An “over-the-shoulder” video recording was made of

participants as they “thought-aloud” while working on the task.

Participants handled office interruptions (phone calls, visitors, etc.) as they normally would.

Study 1: The ResultsMany “keeping” methods were observed:

Send email to self. Send email to others. Print out the web page. Save the web page as a file. Paste URL into a document. Add hyperlink to a web site. Do nothing (and enter URL directly later, search

for or access from another web site). Bookmark the page. Write down the URL on paper.

Survey results (so far)Keeping Method

At least once a week

Do nothing to keep; enter the URL directly. 92.86%

Do nothing to keep; search again to re-access. 78.57%

Do nothing to keep; access via another web site. 71.43%

Make a Bookmark or Favorite 57.14%

Print out the web page 50.00%

Save the web page as a file 35.71%

“Personal Toolbar” or “Links” 35.71%

Send email to self 28.57%

Send email to others 28.57%

Paste the URL into a document 21.43%

Add a hyperlink to a personal web page 14.29%

Use Personal Information Management Software 7.14%

Write down the web address (URL) on paper 7.14%

A Researcher MC is a part-time lecturer and researcher at the

University of Washington. Her task for the observation was to locate web materials that might relate to a lecture she was preparing on the use of Microsoft PowerPoint. MC made frequent use of email. She mailed several URLs to herself – each in a separate email along with comments. On two occasions MC also emailed URLs to colleagues along with comments regarding potential relevance. MC maintained an elaborate organization of folders and subfolders in her email application (Microsoft Outlook) and expressed confidence that she could quickly locate an old email when needed. MC uses Favorites from time to time but declared that “it is a mess” because it hadn’t been organized recently.

A Manager DH is a third-level manager at Boeing. He

travels frequently and is rarely in his office for an hour at a time. DH was interviewed over the telephone. DH rarely accesses the Web directly for workplace matters. When he does use the Web, the task is nearly always limited in time and scope. For example, he may occasionally use the Web to look up contact information for someone or to confirm a flight. DH depends heavily on email – from colleagues, his subordinates and other managers to whom he is responsible in one way or another.

A Functional Analysis

Several functions appear to influence the choice of method: Reminding Context Number of access points – home,

work, road… Ease of maintenance …

A Functional Analysis (cont.)

Por

tabi

lity

Num

ber

of a

cces

s po

ints

Per

sist

ence

Pre

serv

atio

n

Cur

renc

y

Con

text

Rem

indi

ng

Eas

e of

inte

grat

ion

Com

mun

icat

ion

Eas

e of

mai

nten

ance

Email to self Low High Low Med High High High Med Low Med

Email to others Low High Low Med High High Low Low? High High

Print-out High High High Low Low Low High Med High Med

Save as file Med? Low? High High Low Low Low Med? Low Med

Paste URL indocument

Low Low? Low Med High High High? High? Low High

Personal web site Low High Low Med High High High? High Med High?

Search Low High Low Med High Low Low ? Low High

Direct entry Low High Low Med High Low Low ? Low High

Bookmarks Low Low Low Med High Low Low Low Low Low

History Low Low Low Med High Low Low Low? Low ?

A Functional Analysis (cont.)

Por

tabi

lity

Num

ber

of a

cces

s po

ints

Per

sist

ence

Pre

serv

atio

n

Cur

renc

y

Con

text

Rem

indi

ng

Eas

e of

int

egra

tion

Com

mun

icat

ion

Eas

e of

mai

nten

ance

Email to self Low High Low Med High High High Med Low Med

Email to others Low High Low Med High High Low Low? High High

Print-out High High High Low Low Low High Med High Med

Save as file Med? Low? High High Low Low Low Med? Low Med

Paste URL indocument

Low Low? Low Med High High High? High? Low High

Personal web site Low High Low Med High High High? High Med High?

Search Low High Low Med High Low Low ? Low High

Direct entry Low High Low Med High Low Low ? Low High

Bookmarks Low Low Low Med High Low Low Low Low Low

History Low Low Low Med High Low Low Low? Low ?

A Functional Analysis (cont.)

Por

tabi

lity

Num

ber

of a

cces

s po

ints

Per

sist

ence

Pre

serv

atio

n

Cur

renc

y

Con

text

Rem

indi

ng

Eas

e of

int

egra

tion

Com

mun

icat

ion

Eas

e of

mai

nten

ance

Email to self Low High Low Med High High High Med Low Med

Email to others Low High Low Med High High Low Low? High High

Print-out High High High Low Low Low High Med High Med

Save as file Med? Low? High High Low Low Low Med? Low Med

Paste URL indocument

Low Low? Low Med High High High? High? Low High

Personal web site Low High Low Med High High High? High Med High?

Search Low High Low Med High Low Low ? Low High

Direct entry Low High Low Med High Low Low ? Low High

Bookmarks Low Low Low Med High Low Low Low Low Low

History Low Low Low Med High Low Low Low? Low ?

Other Notables Participants seemed to distinguish

between three categories. Web sites used repeatedly – make it

easy to access. Web sites used infrequently but

important to be able to access. Web sites to check out later to see if

useful.

Other Notables (cont.) Some participants went to great lengths to

maintain a single hierarchy. Print web pages to file with other papers. Save email documents to filing system for e-

docs. Work with assistant to establish consistent

organizations across paper documents, e-docs, email & favorites.

Keeping practices appear to vary with a person’s job and relationship to information.

Study 2: Delayed Cued Recall A second study looks at how/how well

people are able to get back to web sites. Session 1: Participant describes each in a set

of web sites they have visited recently – without including name or URL.

Session 2, 3-6 months later: Participants are cued with these descriptions and told to get back to the site as best they can. We observe methods used and problems encountered.

Study 2 Methods Focus on “useful” web sites – sites that

a participant is very likely to want to re-access over the next 12 months (75% or greater).

Sample for different frequencies of access in a typical week. High – daily access in a typical week. Medium – 1 to 3 times in a typical week. Low – not accessed in a typical week.

Measure cue effectiveness. Does the participant recall the task? The site?

Study 2 Results Success rate is high – 90% or better. Roughly 2/3 of the re-finding

methods require no “keeping” forethought. Direct entry of URL Access web site via another web site

(such as a “hub”) Search again

When Participant Recalls Task and Site

Access Freq.

# of Part.

Obj. Success

Subj. Success

Trial time

# of Methods Tried

1st Method Requires “Keeping”

Last Method Requires “Keeping”

Low 11 90% 100% 1:42 1.2 30% 23% Medium 12 98% 100% 1:07 1.2 29% 19% High 11 100% 100% 0:47 1.0 36% 36%

When Participant Recalls Task Only

Access Freq.

# of Part.

Obj. Success

Subj. Success

Trial time

# of Methods Tried

1st Method Requires “Keeping”

Last Method Requires “Keeping”

Low 8 23% 86% 2:59 1.7 42% 52% Medium 3 33% 100% 1:14 1.0 33% 33% High 2 50% 100% 1:59 1.0 50% 50%

Prototyping

… as driven by the data. Simple extensions to Add Favorites to support the following options:

Add a comment. Save Favorite to filing system. Email Favorite.

The Prototype

Conclusions

People use a diversity of methods to organize web information for re-access and re-use.

A functional analysis can help us to understand the diversity of methods observed and their relative popularity.

Conclusions (cont.)

Methods differ in the functions they provide.

No single current method provides all the functions a user may need.

The relative importance of functions (and hence the choice of methods) depends upon the task at hand.

Conclusions (cont.)

A “natural” study of people doing tasks in their workplace can be very useful.

How Can We Do It Better? Better reminding. Better integration. Fewer organizational schemes. Further improvements in “do

nothing” methods. In-line matching for suggested

completions Factor history into search results?

“Stuff I’ve Seen” searching.

For More Information http://kftf.ischool.washington.edu/

Supplemental slides

A Functional Analysis (cont.)

Additional functions: Persistence of information Preservation of information in its

current state Currency of information Ease of integration Communication and information sharing Portability of information

Use of keeping methods by participant group

Res

earc

her

s

Info

rmat

ion

S

pec

iali

sts

Man

ager

s

# of participants 4 8 6

Email to self 75% 25% 0%

Email to others 25% 0% 100% Print-out 50% 13% 17%

Save as file 25% 0% 0%

Paste URL in document 75% 13% 0%

Personal web site 25% 0% 0%

Bookmark (Favorite) 50% 88% 33%

Notes on paper 0% 25% 0%

Links toolbar 0% 13% 17%

Outlook "Note" 0% 13% 0%

A Functional Analysis (cont.)

Por

tabi

lity

Num

ber

of a

cces

s po

ints

Per

sist

ence

Pre

serv

atio

n

Cur

renc

y

Con

text

Rem

indi

ng

Eas

e of

int

egra

tion

Com

mun

icat

ion

Eas

e of

mai

nten

ance

Email to self Low High Low Med High High High Med Low Med

Email to others Low High Low Med High High Low Low? High High

Print-out High High High Low Low Low High Med High Med

Save as file Med? Low? High High Low Low Low Med? Low Med

Paste URL indocument

Low Low? Low Med High High High? High? Low High

Personal web site Low High Low Med High High High? High Med High?

Search Low High Low Med High Low Low ? Low High

Direct entry Low High Low Med High Low Low ? Low High

Bookmarks Low Low Low Med High Low Low Low Low Low

History Low Low Low Med High Low Low Low? Low ?