Transcript ppt - Jaime Teevan

“WHERE’D IT GO?”
Jaime Teevan
Microsoft Research
“WHERE’D IT GO?”
 Study
how people re-find
Difficulties people actually encountered
 Collected via a Web search

 Look
at how people:
Described what they were looking for
 Answered “Where’d it go?”
 Coordinated use with other people

 Explore
ramifications for system design
“WHERE’D IT GO?”
 Example
observation:
I remember when I first joined these forums! There
were little “Did you know” facts about Star Wars at the
front page, but they were replaced with movie quotes!
Why did they disappear?
 Describes
frustration, location, time
 Wants an explanation as an answer
 Facts removed because seeker’s needs
not coordinated with others
UNDERSTANDING RE-FINDING
 Re-finding

60%-80% of Web page visits are re-visits


common on the Web
[Tauscher & Greenberg 1997, Cockburn et al. 2002]
40% of search queries involve re-finding

[Teevan et al. 2007]
 Commonly

“Not being able to return to a visited page”

[GVU 1998]
 Support

a problem on the Web
for re-finding
Search engines, bookmarks, history
RE-FINDING IN A DYNAMIC WEB
 Web

content changes a lot
[Fetterly et al. 2003, Obendorf et al. 2007]
 Re-finding
is harder when things change
Bookmarks break [Hupp & Miller 2007]
 Untrusted repository [Whittaker & Hirshberg 2001]
 Even when not impossible, can make it harder

STUDYING RE-FINDING IS HARD
 Log
analysis
Pro: Naturalistic behavior
 Con: No intent

 Controlled
studies

Need to create re-finding situations

[Bruce, Jones & Dumais 2004]
[Capra & Pérez-Quiñones 2005]

 Can
we get insight into real world
difficulties people encounter?
APPROACH: ANALYZE WEB PAGES
 Gather
Web pages indicative of troubles
Collected via a Web search
 Query: “Where’d it go?”

 Analyze
and code pages
APPROACH: ANALYZE WEB PAGES
 Gather
Web pages indicative of troubles
Collected via a Web search
 Query: “Where’d it go?”

 Other
studies of Web content
Robotic pets [Friedman et al. 2003]
 Injury recovery [Preece 1998]
 Personal data [Good & Krekelberg 2003]

PAGES CONTAINING “WHERE’D IT GO?”
Message boards
Web logs (blogs)
Redirect pages
Articles
FAQs
Other
OVERVIEW OF DATA COLLECTED
Information Target
Format
Total
Web
Digital Other Rhetorical Response
404/Redirect
17
17
0
0
17
0
Web log
21
18
3
0
2
3
Article
23
7
12
4
17
2
FAQ/Help
12
7
4
1
12
0
165
106
54
5
14
113
Other
20
19
1
0
7
1
Total
258
174
74
10
69
119
Message Board
SUMMARY OF FINDINGS
Analyzed data to see how people:
 Described
what they were looking for
 Answered “Where’d it go?”
 Coordinated use with other people
Format
404/Redirect
Web log
Article
FAQ/Help
Message Board
Other
Total
Total
17
21
23
12
165
20
258
Information Target
Web Digital Other
17
18
7
7
106
19
174
0
3
12
4
54
1
74
0
0
4
1
5
0
10
Rhetorical
Response
17
2
17
12
14
7
69
0
3
2
0
113
1
119
DESCRIBING THE MISSING INFO
 Expressions


Ah *pulls out masses of hair* Where'd it go?!?!
Blame self: I think I am going crazy.
 Shared


important
Ok, where’s the link?
Provide info on the demise of .. the .. newspaper.
 Time

context
I noticed it was missing too!
 Path

of frustration
is relative
Recently….
ANSWERING “WHERE’D IT GO?”
 Explanations
I’ve removed the pages I used to have here.
 Maybe Eric didn’t pay his .. hosting fee?
 If Spike doesn’t like .. a post, he’ll take it out.

 Work-arounds

I found it, or something better…
 Resolutions
Most often provided by the changer
 I moved it to the bug reports forum.

COORDINATING MULTIPLE USERS
 People

had different intentions
I think they got removed because there were
only about three of them and they got old fast
 Did
not want others to see information
I was hoping nobody saw it, oops. I got taken
in by that Metallica spoof going around the
net. Found out it was a parody and deleted it.
 Copyright issues

 Wanted
others to see information
DESIGN IMPLICATIONS
 Important
to archive Web content
 Described the missing information

Key into archive by location, relative time
 Answered
“Where’d it go?”
Explanations valuable if archive impossible
 Answers most useful where change occurred

 Coordinated
use with other people
Personalization
 Awareness of what others see

CONCLUSIONS
 Studied
how people re-find on the Web
Difficulties people actually encountered
 Collected via a Web search

 Looked
at how people:
Described what they were looking for
 Answered “Where’d it go?”
 Coordinated use with other people

 Explored
ramifications for system design
FUTURE DIRECTIONS
 Understand
re-finding behavior better
Log analysis of re-visitation
 Analysis of content changes
 Additional studies of re-finding behavior

 Support

Caching systems


re-finding better
Re:Search Engine
Highlighting changes
THANK YOU!
Jaime Teevan, MSR