Propaga'on)based.Social)Aware.Replica'on. for.Social.Video.Contents. - - PowerPoint PPT Presentation

▶

Jun 25, 2023 384 likes •600 views

Propaga'on)based.Social)Aware.Replica'on. for.Social.Video.Contents. Zhi$Wang,$Lifeng$Sun,$Xiangwen$Chen,$Wenwu$Zhu,$$ Jiangchuan$Liu,$Minghua$Chen,$and$Shiqiang$Yang$ Tsinghua)University,)China) The)Chinese)University)of)Hong)Kong)

SLIDE 1

Propaga'on)based.Social)Aware.Replica'on. for.Social.Video.Contents.

Zhi$Wang,$Lifeng$Sun,$Xiangwen$Chen,$Wenwu$Zhu,$$ Jiangchuan$Liu,$Minghua$Chen,$and$Shiqiang$Yang$

ACM Multimedia 2012, Oct. 29 – Nov. 2, Nara, Japan

Tsinghua)University,)China) The)Chinese)University)of)Hong)Kong) Simon)Fraser)University,)Canada)

SLIDE 2

12K$re9shares$ 5K$re9shares$ 2K$re9shares$

SLIDE 3

Social.Network.Changes.How.People.Consume.Videos .

Video& sharing& Over$700$YouTube$video$ links$are$imported$to$TwiEer$ every$minute$ Over$500$years$of$YouTube$ videos$are$watched$every$ day$on$Facebook$

Social$network$is$fundamentally$changing$ video$distribuHon$landscape$

Over$30%.users$select$videos$ using$social$network$service$ in$China$

80%.worldwide.growth$(Q4)2011)@)Q1)2012))

SLIDE 4

Social.Video.Distribu'on.Landscape.Change .

TradiBonal)Video)DistribuBon Social)Video)DistribuBon

Based$on$popularity
Based$on$propagaHon$

Rethinking$distribuHon$for$social$video$contents$

Dynamically$ influenced$by$users$ PropagaHon$in$small$ social$groups$

SLIDE 5

Related.Works .

Cache,.CDN,.and.P2P. [TMM’04,$TMM’07,$MM’09] User.rela'on/influence.for. social.media.distribu'on. [SIGCOMM’10,$INFOCOM’12]

Propaga'on) based.social) aware.replica'on.............. [MM’12]

Content.correla'on.for. UGC.distribu'on. [INFOCOM’09]

SLIDE 6

Replica'on.based.on. social.propaga'on. User) $

RelaHon

Context) $

LocaHon

Content) $

Popularity

PSAR:.Propaga'on)based.Social)Aware.Replica'on.

Propaga'on.inference.

Influence$ Preference$ Locality$ Social$locality$ Geo$locality$ Temporal$locality$

PropagaBon)paKern)

Region$predictor$ Global9audience$ predictor$ Local9audience$ predictor$

PropagaBon)predicBon)

Strategy) Architecture)

Propaga'on.Inference:$Jointly$consider$user,$content,$and$context$

$ $

SLIDE 7

Data)driven.Approach:.Measurement.Study .

Tencent$Weibo$

! 0.5B.users$–$one$of$the$largest$social$network$service$in$China$

$ $ $

A$sample$of$video$sharing$in$20$days$

! 350K.videos$–$5$video$sharing$websites,$14$categories$ ! 1M$users$–$social$relaHon,$behaviors$ ! Over$100$city9level$regions$–$locaHon,$distance$

SLIDE 8

Social.Locality .

PropagaHon$links$are$

short$(average$length$ 1.21)$

Users$are$socially$

close$to$each$other$ How$far$a$video)propagates$along$social$relaHon)

SLIDE 9

Geographical.Locality .

How$many$regions$a$video$propagates$to$

Most$videos$propagate$

in$a$few$regions$$$$$$ (90%$<$5$regions)$

Users$are$located$

nearby$

SLIDE 10

Temporal.Locality .

How$long$a$video$propagates$

Re9shares$happen$in$

the$early$hours$$$$$$$$$ (95%$<$1$day)$

User$behaviors$are$

grouped$around$the$ import$Hme$

SLIDE 11

Architecture.Design.of.PSAR .

PropagaBon)paKern:$social,$geographical$and$temporal$localiHes$ Content$propagaHon$ predicaHon $

Edge9cloud$and$peer9

assisted$architecture$

PropagaHon9based$

social9aware$replicaHon$

SLIDE 12

Region.Predic'on .

Predict$new$regions$a$video$propagates$to$

" Number$of$regions$involved:$ " New$region:$

1000 2000 3000 10 20 30 40 50

Propagation size (x) Number of regions in the propagation (y)

(x, y) (x, 6.5log(0.5682 x))

g(T +1)

v

= c1 log(c2s(T )

v

)

PropagaBon)size) Friends)of)current)users)

SLIDE 13

Global)Audience.Predic'on .

Predict$global$audience$aEracted$

" Global$audience:$

$

10 10

sv (popularity) New users (resource) corrcoef=0.1628

e(T +1)

v

= s(T )

v

(zs(τ (T )

v

)/h(T )

v

)

−4

−2

10 10

sv z(t)/hv (popularity+social) New users (resource) corrcoef=0.6681

Popularity) Temporal)locality) Social)locality)

SLIDE 14

Local)Audience.Predic'on .

Predict$local$friends$aEracted$

" Local$audience:$

−0.2 0.2 0.4 0.6 0.8 1 0.2 0.4 0.6 0.8 1 Preference correlation (T+1 vs. T) CDF

Local)popularity) Friend)preference)

SLIDE 15

Experiment.Setup .

Trace$

! A$sample$of$video$sharing$in$10$days$ ! Guideline:$videos$are$selected$with$diverse$replicaHon$$$$$$

and$replacement$ $

EvaluaHon$metric$

! Local$download$raHo$ ! Local$cache$hit$raHo$

$

SLIDE 16

72%.Users.Locate.Their.Videos.in.2.Hops.in.PSAR .

1 hop 2 hops

thers

20 40 60 80 Number of social hops Fraction of P2P requests (%) PSAR LFU−based Replacement Random Replacement

EffecHveness$of$the$architecture$

2$Hmes$improvement$as$compared$to$ convenHonal$wisdoms$

SLIDE 17

Popular.and.Unpopular.Videos.Effec'vely.Replicated .

10 20 30 40 50 60 70 0.2 0.4 0.6 0.8 1

Number of replications Local download ratio

PSAR Popuarlity−based Random

EffecHveness$of$the$strategy$

4$Hmes$improvement$in$local$fetch$of$ unpopular$videos$$

SLIDE 18

User.Experience.Demo.

SLIDE 19

Summary.

Online$social$network$fundamentally$changes$video$distribuHon$

landscape$

PropagaHon9based$video$distribuHon$by$jointly$considering$user,$

content$and$context$

A$data9driven$approach$unveils$the$propagaHon$paEerns$and$

predicHons$

! General$propagaHon$paEerns$for$architecture$ ! Specific$propagaHon$predicHons$for$strategies$

294$Hmes$improvement$for$local$video$download$

SLIDE 20