1
`
`The Contribution of Thumbnail Image Mou e-o er Te t and
`Spatial Location Memory to Web Page Retrie al in 3D
`
`Mary P. Czerwinski Maarten van Dantzich George Robertson &
`Hunter Hoffman
`ar h On Mi ro oft Way Redmond
`ft R
`98052 SA.
`{ mary z ogr maarten a-hunhof}@micro oft.c m
`
`Mi ro
`
`an mpiri
`val o
`r
`
`·
`
`of th
`
`a ari
`al .
`
`ati n
`
`Ab tract: We
`w
`· on
`red
`that lb
`retu
`
`rt
`·
`
`re ul
`Th
`imp11 ed U
`
`Ki
`
`or ds: D inf: nuati n
`
`· ualizati n. inf. rm ti n retri
`
`fi
`
`Introduction
`1
`nt p per
`In a
`en ·
`d
`d
`ith
`m bani m .
`call th Data
`
`ribed
`d
`t m
`rt
`i
`
`d b p D
`obertS n
`nt
`. ed
`than
`king
`
`r
`ane
`e
`eth
`al
`b.
`bnail miniarur
`.
`patial mm
`
`·
`
`·
`
`ap
`parti
`
`1.1
`
`pa
`info
`Dar
`l 7·
`mp··
`an t
`
`·
`
`.
`lped primariJ)
`·
`docu
`in
`·
`
`ognition
`
`vali
`if th
`
`···
`
`fa
`ap
`naJ
`e of
`and map t
`
`orl
`of th
`land
`h Ip
`In a
`m al
`
`o
`
`· d
`r
`
`· .
`
`.
`
`•
`
`n
`
`.
`
`1
`
`uild a pe
`er tim .
`
`nal organizati n
`
`1
`
`APPLE 1031
`
`

`

`/64
`
`Part 7ko
`
`Tc hni
`
`untain fa ilitate a c mput r
`jnfi
`.
`.
`
`Data
`ural tenden
`in£
`meaningful.
`fir pe
`th m
`
`andl r:
`
`n t
`d o aoization ar
`but organization
`a
`r mem ry . . . all
`'tion
`m
`d
`.
`
`h
`
`)I u
`aid
`th
`
`1.2 Textual Titl
`
`1.3 Clustering in
`lh
`t
`in lin
`
`emory and Retrieval
`pe
`l
`h
`
`i uaJ im
`
`to nhan
`
`2
`
`

`

`ni
`
`o nrriburion of Thumbnail Ima
`
`. Mo u ~o,· r Text and patial Locari n Mm T)' 10
`
`\ b Pa R ri ·al in D
`
`16
`
`un
`
`~ in
`in in lud
`. .
`
`ri
`
`than
`cue
`t p
`tar
`ue cm
`
`.
`
`' th
`b
`
`u,
`
`anti
`
`l
`
`.
`
`·
`
`d
`
`ti
`
`.
`
`.
`
`.
`
`u
`od
`
`rm
`· g t
`p
`b
`
`po e
`return t
`ed
`
`·
`
`.
`g
`that
`u
`·t after a
`i it.
`
`Figure 1: D
`p
`
`Th
`fa illtat
`
`m untain
`
`' th I
`
`2 The Desktop 3D pplication
`D nvironment
`d in thi
`a
`·n and
`th
`n
`taiJ
`t
`o ervi
`urren
`
`, p
`ircuJar laodm
`~
`r b i
`win
`·o
`
`a
`b
`
`lil
`pag
`beth
`environ
`
`pedi
`th
`
`pl
`ouJ
`
`u
`
`.·
`
`n
`mbnail im
`di ml
`
`perfi nn
`aftr6m
`
`u
`
`d
`
`f
`bit
`nail
`
`3
`
`

`

`J
`
`Pan T, vo
`
`1i. clmi al
`
`· ion
`
`3.3 Procedure
`
`th ir old Data.
`
`th
`ed.
`
`n
`to find
`
`ue
`
`Figure 2: am d m uot ·n
`
`ith ot thum nails.
`
`3
`3.1
`
`:ser Study
`objects
`
`··
`
`pre ·
`' in
`turned after
`
`tu
`f
`
`,
`
`•
`·
`re
`
`3.2
`
`timuli and quip.ment
`" h
`d
`m
`
`·
`
`iin the initial
`ri
`
`d
`I
`ed-
`retum
`
`e
`C
`· n th
`
`iumm hia
`d
`17-in b
`· pb Inten
`dand ran
`u ed in
`• · from
`
`l. We pa
`
`retri al tun :
`
`lected pri r t
`
`bi b the parti i pant
`t pag
`ithfo th tw
`
`ipant '
`
`ubj
`
`ti e
`
`ratin
`
`the
`
`d pend nt m
`Tb
`f ubj
`rful indi
`I
`
`po
`
`t be
`ir,m
`
`4
`
`

`

`Tl
`
`nrribution o Tltwrtbnaif Ima • !fou
`
`-o,· r 1i and patial
`
`ati n
`
`to \, b Pa R tri •al in D
`
`167
`
`bl I :
`
`ampl
`
`th
`
`ur u ing
`
`pared
`th injti
`th D
`
`n
`min.
`
`,1
`gth
`4 =
`
`4 Results
`emor
`4.1 Reaction Tim
`Retention after Absence
`
`4.2
`
`umber of Incorrect Retri als
`
`, . 2nd
`
`2nd
`
`11-lncomlct
`
`30
`
`J2s
`120
`i= j 5
`ti 0
`l s
`0
`
`Summary
`Aetrteval CUe
`
`Figure 4:
`bcJi
`findin
`
`f
`
`5
`
`

`

`I
`
`Part 7i
`
`di
`< 1.
`Failed Trials
`
`4.3
`
`and retri val
`
`l
`th
`
`l
`I.bum
`
`th
`and
`
`p
`
`)
`hJ
`11 .
`retri :val u
`
`t
`. The
`ignificanL
`reliabl
`type
`
`4.4
`
`ootributioo of Tbumboai Text
`Title and patial Location
`emory
`
`0
`
`2
`
`...
`
`'
`
`Fi
`tri
`
`retri
`= l
`
`I cim
`tri
`
`nditi n
`
`6
`
`

`

`The Contribution of Thumbnail Ima e, Mouse -over Text and patial Location Memo
`
`to Web Pa Retri al in 3D
`
`169
`
`fi und, F
`
`, 24} = 2.
`
`, p =
`
`al u
`type w
`E = 0.4.
`Subjective Ratings
`.
`b p
`in
`
`.
`
`filled ut
`Figure 2
`u first and
`ne.ral th
`n
`environment.
`re po ·ti: e after
`ificall , ubj
`10 •
`familiar in the
`ling
`o felt that the patial la out
`· on, and
`·
`ormanc 6 mooth
`b page aided
`than in tb
`on. All ignificant
`initi
`·
`pai:recl t-te t. on -
`nnin
`e u ed a one-tailed
`ca
`ubj
`rating
`
`·
`
`ratin
`In
`th
`
`ry
`bje ts e
`month dela e
`
`cond
`of their
`lat r m
`cliff renc
`·
`a
`re J
`o er time.
`In addition, we had
`rank the relati e
`effi
`feature
`in th Data
`ti en
`f fi ar
`· zed might ben fit retrie al
`Mountain that e hypo th
`of the Web pag
`.
`rank of 1 meant that a feature
`th wo
`con id red th
`t for retrie al.
`t
`ranking. On a erage ubj
`thumbnail
`ranked th
`t
`the mo t helpful a g. = 1.8
`folio ed
`a g. = 2.0 and
`the mou -over te t
`clo el
`spatial l atioo
`f the
`eb page on th Data
`th
`Mountain a g.rank = .2 . The pati.alized audio
`feedback
`con · tentl ranked th least helpful cue
`a g.rank = 4.0 . The differen
`ere ignificant
`d termined b a Kru all-Wal.
`X ( ) = 24.02,p < 0.001.
`
`t,
`
`•
`
`The
`LO
`ftware feel familiar.
`The
`to find I.be pa
`I w
`It w
`I king for wilh I.be
`ftware.
`Layin
`fin
`
`*4. 1
`3 6
`
`.67
`
`.56
`
`in
`ebp
`~
`a~~-~
`. Of
`the relati e locao n
`ue ed
`impl
`aid that th
`th nin ubjec
`o
`. The remaining 7 subjec
`h re the Web p
`remembered the general vicini of the
`aid that th
`
`r
`
`p
`
`target pa e after
`o er te t to find the specifi
`
`· ould
`target page.
`
`th m u
`
`-
`
`th
`
`·
`
`5 Conclusion
`We de eloped a 30 en ironmeot for toring fa ourit
`called the Data Mountain. Pr viou
`Web page
`of u ing th
`tudie had b
`n the benefi
`fa r retri .ving
`tored Web pa
`Mountain
`r
`traditional hierarchical tree a er interface . Ho
`unclear fr m th
`hat parti ular
`it
`tudi
`f
`ere beni fiting
`features
`th 30
`· nm
`performance. This t
`o
`c ntribution of
`th
`thumbnail imag
`m u
`-
`er
`th D
`um oail
`u
`able
`r after an
`imag
`.
`ab e
`proxima
`ing
`vi
`pa
`red. Whe
`imag
`.
`ere turned off, performance
`te but onl
`initiaU and onl
`for the thumbnail retrieval cueing
`condition. After 2 block of tri
`b ·
`just
`f
`t at retri
`ing
`· W, b p
`ute1
`no pictorial cu
`.
`· qui
`• en
`r,
`that ubjec ofte
`th
`thumbnail
`their
`preferred cu Of rnt 'l"-'P having th
`thumbnail imag
`o
`bnail ue
`n
`· cult trial type
`bJ
`· rupti n
`·
`umbnail
`u
`ed to use th thumbnail ima
`c uld perform j
`ell using the
`,
`re
`:a.tion
`mou ing-o
`title and
`te
`f th
`page.
`e think that · ·
`rtant finding for the
`ciall for instao
`ign of 3D enviro . .
`d
`in
`h.ich n vi ual fe ·
`· abl
`to di ti:ogui b
`on
`item from another. For
`ample if a ub.
`·
`loolcin
`through a ariety of el
`rroni documeo
`t xt, th re will be littl d" tingu
`g
`ar primaril
`· m repre entati. n.
`vi ual detail in a thumbnail
`Our finding
`u
`at b allowing
`ubj
`t
`to manuall Jay
`d b
`ir data ·
`p
`iding a
`ill be able
`·
`r,
`to rem mber
`red their non-pi
`· al
`if o.ne c
`information quite well. Es
`·
`the
`fa t that ur subj
`n their
`uts
`bad
`fi r a 4-montb period, Lb p
`resul
`·
`at
`thi combination of pictorial im
`· al location
`er te t titl
`and mouse-
`an be
`We
`·
`en urage d
`osider these ·
`ign
`t
`d
`igning information
`.
`de ign i sue
`Future work
`ore th
`·
`in larger, D information vi alization . We
`ere
`
`·
`
`·
`
`7
`
`

`

`170
`
`Pan 1\vo 1l
`
`t
`
`in
`up
`In
`.
`
`i th ut
`.r
`
`la
`
`. F"
`a
`
`exp
`
`·
`
`for th ir
`amin
`ere
`
`.
`
`i
`
`.
`
`mm
`tinoe to
`·gos
`d
`uld
`we
`.th and
`in rder
`arti uJar
`inn
`d
`ill
`e
`improvin e e of
`
`G
`
`• J
`I
`ni
`
`• •
`g
`6,
`
`ilitati
`nd
`.
`
`Jon
`
`, C. B. 19 6, S 1stemati Softwa~ Dev lopment sin
`VDM. Prentii Hall lotemati naJ .
`
`etapb
`fere
`11
`
`r aniuu:i
`pen e
`MotivaJi.on
`
`M
`Th
`l.
`
`in K.
`lrol
`·c
`
`Pick. H. L.
`,
`erti aJJ -Alligned
`f
`11111 nr and Behavior 2 , 4
`
`·'Im grating
`e
`
`Pau n B.
`
`Hi t ry
`
`f
`
`em ry
`
`rts",
`
`Acknowledgements
`ank u Dumai
`1 man fi r b Jp
`and the
`er interf
`fi r their
`r
`n th Data
`n.
`
`References
`B u eld,
`in th
`
`Bo
`
`Darleen. R.
`an B
`eer
`H
`pp.L4 -
`
`Dar
`
`M.
`bi
`ord
`·io
`
`a1
`a
`
`r
`
`Re all
`amin
`
`trategi
`.
`, in G. van der
`of CH1 '96:
`M
`
`R be
`
`TuJ
`
`n
`B
`
`g
`4
`
`. J.
`J
`in
`g
`rimental I
`al
`peri
`7.
`
`D . M .
`p
`usin
`· gatio
`Exp rim ntal
`
`rganizati o in Fre Recall
`69, 44-
`Ii logi al R
`i
`
`nderwood, B . J.
`. ( I
`hulz. R .
`and erbal Leamin , Llppin
`tl.
`
`, Mem1ingfulne
`
`aller, D. Hunt.
`patiaJ Koo
`Pre ence 7.
`
`app, D .
`e in
`.
`
`• ·'Tb Transfer f
`
`8
`
`