20130106_social_computing

advertisement
网络专题选讲
华中科技大学 电子与信息工程系
程文青
chengwq@mail.hust.edu.cn
2013年1月
社会网络应用专题选讲
华中科技大学 电子与信息工程系
互联网技术与工程研究中心
黑晓军
Email: heixj@hust.edu.cn
Web: http://itec.hust.edu.cn/~heixj
2013.1
Outline


Introduction
Case study



《网络专题选讲 》
Traffic transport
NetTube: Exploring Social Networks for
Peer-to-Peer Short Video Sharing, 2009
Incentive
P2P Trading in Social Networks: The Value
of Staying Connected, 2010
Recommendation
Circle-based Recommendation in Online
Social Networks, 2012
-3-
Internet Topology
Introduction
我们生活在一个关系的社会
5
社会网络应用
6
Friend network in Facebook
7
Co-authorship network
8
Co-authorship in network science
9
Ingredient networks
10
911事件——犯罪网络
11
Social Networking(General public)
Social Networking(General public)
Social Networking(Academia)
专著
专著
研究现状

主要研究机构



国外:MIT、Stanford、Maryland、USC、HP、Michigan
国内:IBM中国研究院、微软亚洲研究院、中科院、中国传媒大学
、清华大学、南京大学
近年来社会网络成为国内外研究热点







美国国家科学基金会(NSF)将社会计算研究领域提供专项资金(2010)
美国计算机协会ACM,Workshop on Social Network Mining and
Analysis(2007~2012)
WWW会议成立“Social Network and Web2.0 Track”论坛(2009)
SIGCOMM,ACM SIGCOMM Workshop on Online Social
Networks(2009~2012)
EuroSys, Workshop on Social Network System(2009~2012)
互联网测量会议(IMC),海量数据仓库国际会议(VLDB),信息与知识管理
(CIKM)大量关于社交网络文章
全国网络科学论坛(2004~2012),全国复杂网络会议(2005~2011)
17
NetTube:
Exploring Social Networks for
Peer-to-Peer Short Video
Sharing
IEEE INFOCOM, 2009
Xu Cheng and Jiangchuan Liu
School of Computing Science
Simon Fraser University
British Columbia, Canada
October 2009
Background (1)

Social Networked Media Sharing – new killer
Internet application
 Since 2005
 Rich user-generated content (UGC) sharing
 Social networks



Among users
Among videos
Changing the popular culture
Background (2)

YouTube – a representative
 Popular





Market share of around 43%
More than six billion videos viewed in January 2009
Consumed as much bandwidth as the entire Internet in
2000
3rd visits among all Internet sites (after Google and Yahoo)
Fast growing


20% growth rate per month
15 hours of new videos are uploaded
every minute
Motivation (1)

The YouTube Crisis – all other sites’ challenge
 Severely hindered by client/server architecture
 Bandwidth costs




Consumed as much bandwidth as the entire Internet in
2000
$1 million a day for server bandwidth!
Sold to Google for $1.65 billion in Nov. 2006
Performance and scalability

“Slow” among the surveyed sites by Alexa.com
Motivation (2)

Peer-to-peer (P2P) – alternative to Client/Server
 New generation of communication paradigm


Scale well with larger user base


Each peer contributes its bandwidth to serve others
More users, more resources contributed
Success already seen in



BitTorrent, eMule, eDonkey (file sharing)
Video broadcasting
…
P2P架构

没有永远在线的服务器

任意主机可以同另一个主
机进行通信
peer-peer
节点可以间歇性的连入系
统,IP地址可能会变化

23
对等网络流媒体系统

两大设计空间



如何形成重叠网络?
如何传输内容?
现有体系结构


树状拓扑 + 推式内容传输
 ESM, Yoid, CoopNet, SplitStream, Bullet, Chunkspread
…
网状拓扑 + 拉式内容传输
《网络专题选讲》
-24-
网状-拉式对等网络流媒体系统

这类系统非常类似于BitTorrent
《网络专题选讲》
-25-
节点软件结构

双缓存



《网络专题选讲》
积极下载 vs 保守下载
处理丢失的数据块
缓存控制
-26-
缓存映像(buffer map)


缓存映像反映了节点缓存所拥有的数据块信息
此映像可以被用来评估用户的播放质量
《网络专题选讲》
-27-
对等网络中的问题



内容组织和搜索
内容传输
信誉、激励及安全相关问题
28
CoolStreaming


The first practical large-scale P2P NetTV
Origin of data-driven mesh design
 With many follow-ups: PPLive, PPStream, UUSee
…
X. Zhang, J. Liu, B. Li, and T.-S. P. Yum, CoolStreaming/DONet: A Datadriven Overlay Network for Live Media Streaming, IEEE INFOCOM'05,
March 2005. >800 citations
J. Liu, S. G. Rao, B. Li, and H. Zhang, Opportunities and Challenges of Peerto-Peer Internet Video Broadcast, Proceedings of the IEEE, Vol. 96, No. 1, pp.
11-24, January 2008.
Data-Driven Mesh

Core operations



Easy to implement


no need to construct and
maintain a complex global structure
Efficient


Every node periodically exchanges data availability
information with a set of partners
Then retrieves unavailable data from one or more partners,
or supplies available data to partners
data forwarding is dynamically
determined according to data availability
Robust and resilient

adaptive and quick switching among multi-suppliers
Challenges and Opportunities (1)

Challenges – Drastically different statistics
 1.5 year measurement of 5 million videos


http://netsg.cs.sfu.ca/youtubedata/
Short video clips – stability


99.6% are less than 700 seconds
“I don’t want to wait for 30 seconds for
a two-minute video!”



Searching for sources
High churn rate: join/leave system
Huge number of videos – scalability



Highly skewed
Inefficient for unpopular videos
Very few users watch the same one
Challenges and Opportunities (2)

Opportunities – Social networks
 No longer independent – videos have related videos
 Small-world – strong clustering
 Important role
NetTube Design (1)

Bi-layer overlay network
 Lower-layer – per-video


Download and uploading
Peers stay in previous overlays as
sources


Larger and more stable
Upper-layer – social network



Connected by the same peers in
different lower-layer overlays
Conceptual relation for searching
Social network brings similar peers
closer


Clustering
Efficient searching
NetTube Design (2)

Bloom filter based indexing
 An efficient approach to keep track of peers’
cached videos
 Bloom filter



An m-bit array using k hash functions
Space-efficient
Scalable indexing table



Fast searching
Table size is scalable with the number of videos
Search locally and search in the upper-layer overlay

Social network clustering the similar video
NetTube Design (3)
Transmission scheduling:
From which partner to fetch which data segment ?

Constraints
 Data availability
Playback deadline
 Heterogeneous partner bandwidth
Rarest-first (BitTorrent’s) doesn’t work !


NetTube Design (3)


Variation of Parallel machine scheduling
 NP-hard
Conventional Heuristics
 Message exchanged




Window-based buffer map (BM): Data availability
Segment request (piggyback by BM)
Less suppliers first
Multi-supplier: Highest bandwidth within deadline first
NetTube Design (3)

Short video ?

CODAS: Collaborative Delay-Aware Scheduling
NetTube Design (4)

Social network assisted pre-fetching
 Most peers finish downloading before playback
ends
- free time available (about 80 seconds on average)

Using free time to reduce startup delay

Prefix pre-fetching



Multiple pre-fetching



Avoid wasting bandwidth and space
Enable multiple pre-fetching
Accuracy increases greatly
Accuracy increases as watch more videos
Pre-fetching among neighbors


Easy to implement
Social network helps improve efficiency
Performance Evaluation (1)

Simulation
 Configuration




Based on about 7,000 crawled videos
Scale to more than 10,000 heterogeneous clients
Compare with PA-VoD (MSN Video)
Bandwidth reduction


Save significantly more
More scalable
Performance Evaluation (2)

Simulation
 Impact of social network


Find more sources: more than 95% within 2 hops
Greatly increase pre-fetching accuracy
Performance Evaluation (3)

PlanetLab experiment
 Configuration


Maximum 235 PlanetLab nodes
Experiment results



Server bandwidth reduction: more than 40%
Startup delay: average 2.2 s
Playback continuity
Summary

Contribution



Techniques





First social network assisted P2P system for short video
sharing
IWQoS’08, INFOCOM’09, IEEE Transactions on Multimedia
Bi-layer overlay network
Bloom filter based indexing
Social network assisted pre-fetching
Collaborative delay-aware scheduling
Evaluation results

Greatly reduce server bandwidth



Much lower maintenance cost: $1 million → $60 K
Inherently scalable – P2P
Greatly reduce playback delay


Satisfying startup delay
Continuous playback
P2P Trading in Social Networks:
The Value of Staying Connected
IEEE INFOCOM, 2010
Zhengye Liu, Hao Hu, Yong Liu, Keith Ross, Yao Wang, and Markus
Mobius
Polytechnic Institute of NYU
Dept. of Economics, Harvard Unviversity
43
Outline

Background: P2P Incentive

Networked Asynchronous Bilateral Trading (NABT)

NABT Efficiency Theory

NABT Simulations

Conclusions
44
P2P Apps: BitTorrent
45
P2P Apps: Skype
46
Peer-Assisted Video Streaming

Large scale deployments on
Internet



Leading P2P Video Companies





47
thousands of live/on-demand channels
millions of world-wide users daily
CoolStreaming
PPStream
PPLive
Sopcast
UUSee
1
6
2
5
3
4
Major P2P Issues

Traffic localization


Security




48
P4P
Attacks on
Attacks from
Lack of uniform API
Incentives for peers to contribute resources
Partially Successful P2P Incentive

BitTorrent is popular




50+ client implementations
Dozen public trackers
5-10 million users
Why BitTorrent?
P2P design


Incentives

Resources
First generation P2P applications: Gnutella

49
+
70% of users are free-riders
Second generation P2P applications: BitTorrent
The BitTorrent Incentive
To get files faster…
contribute more bandwidth
Implementation of incentive:

50
The rich play/trade with the rich
BitTorrent: Tit-for-tat
(0) Everyone nominally has four trading partners
(1) Alice tries sending to Bob. Is he rich?
(2) Alice becomes one of Bob’s top-four providers; Bob reciprocates.
(3) Bob becomes one of Alice’s top-four providers.
Tit-for-Tat: Live P2P Video
To get better video quality…
contribute more bandwidth
Layer 3
LC31
LC32
LC33
LC34
Layer 2
LC21
LC22
LC23
LC24
Layer 1
LC11
LC12
LC13
LC14
“LayerP2P: Using Layered Video Chunks in P2P Live Streaming”,
Z. Liu, Y. Shen, K.W. Ross, S. Panwar, Y. Wang,
IEEE Transactions on Multimedia, November 2009.
52
“Substream Trading: Towards an Open P2P Live Streaming System”,
Z. Liu, Y. Shen, K.W. Ross, S. Panwar, Y. Wang,
Inter Conf on Network Protocols (ICNP), October 2008
Limitations of Tit-for-Tat

Tit-for-Tat is synchronous trading


Tit-for-Tat == Barter (物物交换) in primitive
economy
Barter is highly inefficient



53
Alice and Bob can trade if and only if they
simultaneously have data for each other in a
short time period
fails if lack of “double coincidence of wants”
failure example:

Tit-for-tat does not provide incentive for seeding
Currency-based Trading

Currency improves trading efficiency in
modern economy

Asynchronous trading regulated by money

54

users accumulate for providing services
and later spend for acquiring services
Major Issues/Solutions
Solutions:
 Cheating
 Counterfeit
 Dispute Resolution
55
Banking System
Market Regulation
Trading Policy
Court System
Law Enforcement
……
Global Currency in P2P?

Peers trade with each other using digital cash



Heavyweight coordination infrastructure needed



56
earn cash by contributing resources to provide services
to other peers,
pay cash to consume services provided by other peer.
banking/regulation/court/enforcement
hard to justify for P2P trading goods carrying low value.
Only limited research attempts, no large-scale
deployment
Desirable P2P Incentive Mechanism

High Trading Efficiency




Cheating-proof



isolate and punish cheaters
prohibit collusions
Low-degree of Coordination


57
trade asynchronously
trade with many peers
trade diverse set of goods/services
light-weight and distributed protocols
low management cost
Outline

Background: P2P Incentive

Networked Asynchronous Bilateral Trading (NABT)

NABT Efficiency Theory

NABT Simulations

Conclusions
58
Alternative Trading Systems in Social Networks

Asynchronous Trading



Networked Trading


59
exploit trust between friends
allow debt: providing a service without immediate payment
exploit trust in network of friends
trade with indirect friends
“Trust and social collateral”.
Dean Karlan, Markus Mobius, Tanya Rosenblat, and Adam Szeidl.
Quarterly Journal of Economics, 2008.
Friendship as Trading Collateral (抵押)
Resolve cheating/disputes:
Terminate friendship!
60
P2P Trading in Social Networks

Networked Asynchronous Bilateral Trading
(NABT)





61
Social network: peers belong
to an underlying social network
Pair-wise credit: friends
maintain pair-wise credits
Asynchronous trading: peers
can use their credits anytime they want
Credit limit: each peer sets a credit limit for each of
its friends
Networked trading: peer trades with a remote peer
by transferring credits through a chain of friends
links.
Async Trading Between Direct Friends

A pair of friends maintain local credit
balance

bij = amount of credits b + ∆
that j owes i
bij=-bji
Alice
update balance
upon services
AB



Control risk of defaulting


62
bBA - ∆
Cij = credit limit for j set by i
- Cji ≤ bij ≤ Cij
incentivizes users
Bob
Networked Trading via Intermediaries

To access service on a remote peer
1.
2.
3.
4.
find a path of friend links in social network
arrange a series of credit transfers along path
intermediaries update credit balances with
upstream and downstream friends, and break
even
remote peer provides b +∆
b -∆
, bBA-∆, b +∆
requested service
CB
AB
BC
Alice
Bob
63
Charlie
NABT Issues
NABT is decentralized, and effective for
resolving disputes.
But
 Is NABT efficient?


How to set credit limits Cij ?

Can users free-ride in NABT?
64
Outline

Background: P2P Incentive

Networked Asynchronous Bilateral Trading (NABT)

NABT Efficiency Theory

NABT Simulations

Conclusions
65
NABT Efficiency

Single trade can be exercised if and only if a
credit transfer can be arranged



Multiple trades coupled through the
underlying social network


66
subject to social network connectivity
obey credit limit on each social link
later trades work with credit balance resulted
from earlier trades
concurrent trades compete for credit transfer
NABT Efficiency Model

Given:



underlying social network:
credit limits as link weights:
service demand matrix:
: cost charged by user k to serve user l.


Find

credit transfer flows for all demands



resulting credit balance bounded by credit limits

67
: credit flow for demand d on social link <i,j>
credit flow conservation on intermediaries
NABT Credit Flow Routing

Similar to classical network flow problem, but:



credit balance on link can be negative
credit flows in opposite directions cancel
Example:

Circular Service Demands: A wants a file on B, B
wants a file on C, and C wants a file on A
B
bAC=1 A
bBA=1
B
C bCB=1
credit routing scheme 1
68
bAC=0 A
bBA=0
C bCB=0
credit routing scheme 2
Balanced Demand

For each user k, total service he provides
(regardless of receivers) equals total service
received (regardless of providers)

Theorem 1: Any balanced demand can be
executed as long as users involved in the
demand sets are connected.
NABT is as efficient as global currency


69
networked Tit-for-Tat: peers play tit-for-tat with
whole network instead of another peer
Unbalanced Demand

For at least one user, service contribution
does not equal to service consumption.




net-service contribution:
service sources:
service sinks:
aggregate net-service imbalance

70
Extended Social Network


augment social network with a virtual source,
a virtual sink, virtual links
Example
aggregate
net-service
imbalance
71
Efficiency with Unbalanced Demand


Theorem: An unbalanced demand is executable iff
the min-cut between the source s+ and sink s- in
extended social network is greater than or equal to
the aggregate net-service imbalance.
What matters:




What does not matter:

72
underlying social network topology
credit limits on social links
service imbalance between a user
and whole network
service imbalance between individual pairs of users
Dynamic Payment Routing
73

Time is slotted

Demands are now sequential H(1), H(2),…

Suppose we succeed at executing H(1),…,
H(k-1).

Theorem: To successfully execute H(k), we
do not have to worry about how we executed
H(1),…,H(k-1).
Outline

Background: P2P Incentive

Networked Asynchronous Bilateral Trading (NABT)

NABT Efficiency Theory

NABT Simulations

Conclusions
74
Preliminary NABT Protocol Design

On-demand credit flow routing






Dynamic credit-limit setting


75
locate service providers
send out credit-transfer request through controlled flooding
request propagates along friends links with enough credit
space
When request hits one providers, it sends back reply
through reverse path to establish credit transfer on
intermediaries.
Complete credit transfer and service
increase credit-limit linearly after each fulfilled transaction
decrease credit-limit multiplicatively after each
unfulfilled/disputed transaction
Simulation Study

Trading with global currency (GCT):




Synchronous Trading (ST):



Two peers can trade if and only if they can supply files to each
other simultaneously
If peer i downloads a file from peer j, peer j will download a file
from peer i.
Two-hop NABT:



76
Global currency and a centralized bank
Each peer has Bi initial credits and each file costs one credit
If peer i downloads a file from peer j, peer i pays 1 credit to peer j
Peers are connected in an underlying social network
A requesting peer requests files from its friends (one-hop friends)
and the friends of its friends (two-hop friends)
If peer i downloads a file from peer j within two hops, peer i passes
1 credit to peer j
Simulation Setup

Peer profile






File profile


77
Social network with a topology collected from MySpace
Totally 10,000 peers
Peer upload bandwidth
 37% Ethernet users (1.2Mbps) + 63% residential users
(400 kbps)
Willingness for sharing
 10% content-rich peers (1,000 files) + 90% content-scarce
peers (50 files)
Online and offline
 Markov ON-OFF process (On time = Off time = 12 hours)

Totally 10,000 different files
Files are small and have the same size of 3MB
File popularity follows a Zipf distribution
Trading Efficiency

Request success ratio: The ratio of fulfilled requests to
the total number of requests
78
CDF of request success ratio
Importance of Trading Intermediaries
79
CDF of request success ratio for the systems with and without
intermediaries
Service Differentiation of NABT
80
Relation between request success ratio and upload
contribution (in terms of number of uploaded files)
Conclusion

NABT -- a new P2P trading paradigm over
social networks



NABT is efficient




81
exploits trust between friends, and friends
network
trade asynchronously, and over network,
light-weight, distributed

almost as efficient as global currencies
support networked tit-for-tat
topology and credit limits matters
memoryless
Open Research Issues




incentives for intermediaries
isolate and punish cheators
dynamic credit-limit setting
heterogeneous NABT market




82
diverse set of services
exchange ratio between pair-wise credits
deal-making
……
Take Away Messages

Asynchronous incentives are critical
for taking P2P to the next level

Async incentives require money

The future of P2P may lie in social
networks
83
Circle-based Recommendation in Online Social
Networks
ACM KDD 2012
Xiwang Yang, Harald Steck*, and Yong Liu
Polytechnic Institute of NYU
* Bell Labs/Netflix
84
Outline

Background & Motivation

Circle-based RS





85
Trust Circle Inference
Trust Value Assignment
Model Training
Evaluation
Conclusion & Future work
Social Recommenders Everywhere
86
Collaborative Filtering (CF)

Most Used and Well Known Approach for
Recommendation

Finds Users with Similar Interests to the
target User

Aggregating their opinions to make a
recommendation.
87
User Based Collaborative Filtering
Target
Customer
Aggregator 
u
w u  ru , i

Prediction
88
u
wu
Item-based Collaborative Filtering
89
Item-Item Collaborative Filtering
Aggregator
90

i
w i  ru , i

i
Prediction
wi
Matrix Factorization (BaseMF) [NIPS08]
 Introduced by R. Salakhutdinov
and A. Mnih Probabilistic matrix
factorization. In NIPS 2008
 Model based approach
 Latent features for users
QR
u0  d
 Latent features for items
i d
PR0
91
P and Q have normal priors
Matrix Factorization (BaseMF)
 Prediction Model
T
Rˆ  rm  Q P
P (R | P, Q , ) 
2
R
  [N ( R
u ,i
2
| Rˆ u , i ),  R ]
R
I u ,i
all u all i

  [N ( R
| rm  Q u Pi ),  ]
T
u ,i
2
R
R
I u ,i
all u all i
 Objective Function
1
2
92

( u , i ) obs .
( R u ,i

2
2
2
ˆ
 R u , i )  (|| P || F  || Q || F )
2
P and Q have normal priors
Related Work-Social Recommender

Social Recommendation (SoRec) Model



Social Trust Ensemble (STE) Model





93
SIGIR’09
User’s rating influenced by social friends
SocialMF Model


CIKM’08
Factorizing social trust matrix together with user rating
matrix
RecSys’10
User’s latent feature (taste) influenced by social friends
Handle trust propagation in social network
Using whole trust network for item rating prediction
SocialMF [RecSys2010]


Social Influence  behavior of a user u is affected by
his direct neighbors Fu.
Latent factor of a user depend on his neighbors.
*
 S u ,v


is the normalized trust value.
Prediction Model:
Objective:
1
2

2
( R u , i  Rˆ u , i )
( u , i ) obs .


*
*
T 


 ( Q u   S u , v Q v )( Q u   S u , v Q v ) 
2 all u 
v
v



2
94
(|| P || F  || Q || F )
2
2
Proposed Improvements for Current
Social Recommender

Social networks include multiple circles



A more refined social trust information—richer information
Incorporate circle information in Social Recommender
Use trust circles specific to an item category when predict
rating in this category

95
e.g. Trust Circle of “Music”, Trust Circle of “Cars”, etc
Proposed Improvements for Current
Social Recommender
Existing circles (Google+, Facebook) not corresponding
to an item category
96
Proposed Improvements for Current
Social Recommender




97
In existing multi-category rating datasets, no circle information
User trusts different subsets of friends in different domains (Cars,
Music…)
User trusts different friends differently, related to friend’s
expertise value
Should use trust circle specific to item category
Outline

Background & Motivation

Circle-based RS





98
Trust Circle Inference
Trust Value Assignment
Model Training
Evaluation
Conclusion & Future work
Trust Circle Inference

User v is in inferred circle c of u iff u trust v in original social
network and both of them have rating in category c
Original Social Network
Inferred circle
for category C1
99
Inferred circle
for category C2
Inferred circle
for category C3
Outline

Background & Motivation

Circle-based RS





100
Trust Circle Inference
Trust Value Assignment
Model Training
Evaluation
Conclusion & Future work
Trust Value Assignment

101
CircleCon1: Equal Trust
Trust Value Assignment

CircleCon2: Expertise-based Trust

102
assign a higher trust value or weight to the friends that are
experts in the circle / category.
CircleCon2: Expertise-based Trust

Variant a:

103
Expertise based on number of ratings in a circle
CircleCon2: Expertise-based Trust

Variant b:
(c)
Ev
 N v  v
(c)
(c)
Dw records the proportions of
ratings user w assigned in all
categories. It reflects the
interest distribution of w cross
all categories
104
CircleCon3: Trust Splitting
Original trust link




105
trust link in c1
trust link in c2
Trust due to followee’s rating in one category
Likelihood u2 trusts u1 in C1, C2 ?
Infer likelihood proportional on u2’s number of ratings in C1 and C2.
Assign trust value in a category proportional to the likelihood u2 trusts
u1 in a category
CircleCon3: Trust Splitting
N

106
c1
u1
 9, N
c2
u1
1
S u 2 1, u1  0.9, S u 2 2, u1  0.1
(c )
(c )
Normalize across followees
Outline

Background & Motivation

Circle-based RS





107
Trust Circle Inference
Trust Value Assignment
Model Training
Evaluation
Conclusion & Future work
Model Training
Training with ratings from each category

Predict user’s rating in category c
Input rating: rating in category c
Input social network: Circle c



(c)
(c)
(c)
( c )T
Rˆ u , i  rm  Q u Pi
L
(c )
1
2

(R

,Q
(R
(c )
(c)
u ,i
,P
(c )
,S
) 
( c )*
  is social information
weight
(c) 2
 Rˆ u , i )
( u ,i ) o b s .

2

(c )

2

(c)
 (Qu 
u 

a ll
S
( c )*
u ,v
(c)
Qv
v
(c)
)( Q u

S
( c )*
u ,v
v
T 
) 

(|| P
(c)
||
2
F
 || Q
(c)
2
F
|| )
i
(c)
0 d
u0  d
,Q  R
P R
i0( c ) is the number of
items in category c
(c)
Solved by gradient descent
108
(c)
Qv
(c)
Model Training

109
Training with ratings from each category
Model Training

Training with ratings for all categories
Predict user’s rating in category c
Input rating: rating from all categories
Input social network: Circle c



L
(c )
1
2


(R,Q

(c )
110
(c )
( c )*
,S
) 
2
( R u , i  Rˆ u , i )
( u ,i ) o b s .


(c)
(
Q
 u 
2 a ll u 

(|| P
(c)
2
P
,P
(c)
2
,Q
(c)
v
(c)
(c)
)( Q u


v
|| F  || Q
i d
R0

( c )*
S u ,v Q v
(c)
2
|| F )
R
u0  d
( c )*
(c)
S u ,v Q v
T 
) 

Outline

Background & Motivation

Circle-based RS





111
Trust Circle Inference
Trust Value Assignment
Model Training
Evaluation
Conclusion & Future work
Epinions Data
112
Performance Metrics

RM SE 
( u , i ) R test
2
( R u , i  Rˆ u , i )
| R test |
M AE 

( u , i ) R test
| R u , i  Rˆ u , i |
| R test |
113
Training with per-category ratings
114
Training with per-category ratings
L
(c )
(R
(c )
,Q
(c )
,P
(c )
,S
( c )*
) 

(c)
(c) 2
(c)
2
(c)
2
( R u , i  Rˆ u , i ) 
(|| P
|| F  || Q
|| F )
2 ( u ,i ) o b s .
2
1



2
115

(c)
 (Qu 
u 

a ll

v
( c )*
(c)
S u ,v Q v
(c)
)( Q u


v
( c )*
(c)
S u ,v Q v
T 
) 

Training with ratings from all categories
116
CircleCon3 of training with
per-category rating
Training with ratings from all categories
117
Training with ratings from all categories
118
Summary

Propose a novel Circle-based Social Recommendation
framework



Split original social network to different circles, one circle
corresponding to one item category
User trusts different subsets of friends in different domains(Cars,
Music…)
User trusts different friends differently, based on friend’s expertise

Outperforms the state-of-the-art social collaborative filtering
algorithms

Show the promising future of circle-construction techniques in
Social Recommender
119
小结
Social networking has been changing
the way which people communicate!
120
Reading List

Lada Adamic, Social Network Analysis, https://class.coursera.org/sna-2012001/wiki/view?page=syllabus

World By David Easley and Jon Kleinberg, Networks, Crowds, and Markets
Reasoning About a Highly Connected, Cambridge University Press, 2010
http://www.cs.cornell.edu/home/kleinber/networks-book/

Xu Cheng and Jiangchuan Liu, "NetTube: Exploring Social Networks for
Peer-to-Peer Short Video Sharing", IEEE INFOCOM, 2009.

Zhengye Liu, Hao Hu, Yong Liu, Keith Ross, Yao Wang, and Markus
Mobius, “P2P Trading in Social Networks: The Value of Staying Connected”,
in the Proceedings of IEEE Conference on Computer and Communications
IEEE INFOCOM, 2010

Xiwang Yang, Harald Steck and Yong Liu, “Circle-based Recommendation
in Online Social Networks ”, in the Proceedings of ACM SIGKDD
Conference on Knowledge Discovery and Data Mining (KDD 2012), Long
Paper, August, 2012
-121-
Download