social network something interesting ruizhi gao. contents the born of social networks new types of...

45
Social Network Something Interesting Ruizhi Gao

Upload: samara-beavers

Post on 15-Dec-2015

218 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Social Network Something Interesting Ruizhi Gao. Contents The Born of Social Networks New types of Social Networks My Social Networks Research related

Social Network Something Interesting

Ruizhi Gao

Page 2: Social Network Something Interesting Ruizhi Gao. Contents The Born of Social Networks New types of Social Networks My Social Networks Research related

Contents

• The Born of Social Networks• New types of Social Networks• My Social Networks• Research related

Page 3: Social Network Something Interesting Ruizhi Gao. Contents The Born of Social Networks New types of Social Networks My Social Networks Research related

History

Page 4: Social Network Something Interesting Ruizhi Gao. Contents The Born of Social Networks New types of Social Networks My Social Networks Research related

Early Years

AIM

ICQ

They are client-based application which can allow you to add friends and have friends list. You can communicate with each other by send text message, image and other information

Problems: You have to remember the account ID if you want to add new friends. Your friends list is not visible It’s more like a communication tool but SNS

Page 5: Social Network Something Interesting Ruizhi Gao. Contents The Born of Social Networks New types of Social Networks My Social Networks Research related

SixDrgree.comCombine the function of many different tools like ICQ and AIM, and provide a new search approach based on users’ own information.

SixDegrees.com was a social network service website that lasted from 1997 to 2001 and was based on the Web of Contacts model of social networking. It was named after the six degrees of separation concept and allowed users to list friends, family members and acquaintances both on the site and externally; external contacts were invited to join the site. Users could send messages and post bulletin board items to people in their first, second, and third degrees, and see their connection to any other user on the site. It was one of the first manifestations of social networking websites in the format now seen today. Six Degrees was followed by more successful social networking sites based on the "social-circles network model" such as Friendster, MySpace, LinkedIn, XING and Facebook. ----- Wikipedia http://en.wikipedia.org/wiki/SixDegrees.com

Page 6: Social Network Something Interesting Ruizhi Gao. Contents The Born of Social Networks New types of Social Networks My Social Networks Research related

LiveJournal.com

It gives a new way that people mark others as Friends to follow their journals and manage privacy settings

Page 7: Social Network Something Interesting Ruizhi Gao. Contents The Born of Social Networks New types of Social Networks My Social Networks Research related

SNS on business networks

• Ryze• Tribe.net• Friendster

Page 8: Social Network Something Interesting Ruizhi Gao. Contents The Born of Social Networks New types of Social Networks My Social Networks Research related

MySpace

• Myspace could grow rapidly because many other SNS are trying to collect fees.• MySpace open some public webpages to famous bands or singers so their fans can

follow. • MySpace did not restrict users from add HTML code into the forms to make their

own pages special.

Page 9: Social Network Something Interesting Ruizhi Gao. Contents The Born of Social Networks New types of Social Networks My Social Networks Research related

Today

Twitter is an online social networking service and microblogging service that enables its users to send and read text-based messages of up to 140 characters, known as "tweets"

Harvard - Only High School Networks Corporate Networks

Page 10: Social Network Something Interesting Ruizhi Gao. Contents The Born of Social Networks New types of Social Networks My Social Networks Research related

Reference• Wikipedia• Boyd, d. m., & Ellison, N. B. (2007). Social network

sites: Definition, history, and scholarship. Journal of Computer-Mediated Communication, 13(1), article 11.

• Visual Academy http://www.onlineschools.org/visual-academy/history-of-social-networking/

Page 11: Social Network Something Interesting Ruizhi Gao. Contents The Born of Social Networks New types of Social Networks My Social Networks Research related

New

Page 12: Social Network Something Interesting Ruizhi Gao. Contents The Born of Social Networks New types of Social Networks My Social Networks Research related

Types of Social Networks

• General Facebook, Twitter … • School, college Classmates.com, Friends Reunited(UK)…• Art Community deviantART, Taltopia• Movies , TV series Youtube, douban, Flixster, Filmow … • Photo sharing Flickr, Fotki, DailyBooth … • …

Page 13: Social Network Something Interesting Ruizhi Gao. Contents The Born of Social Networks New types of Social Networks My Social Networks Research related

SoLoMo

• SoLoMo, short for social-local-mobile, refers to a more mobile-centric version of the addition of local entries to search engine results.(GPS)

• SoLoMo based social network. The most famous one is Foursquare

Page 14: Social Network Something Interesting Ruizhi Gao. Contents The Born of Social Networks New types of Social Networks My Social Networks Research related

FoursquareYou can check-in in different place (restaurant, movie center … ) and leave your comments. The more you checked in, more point you earn. You also can be the “mayor of one place” if you check-in many times

Page 15: Social Network Something Interesting Ruizhi Gao. Contents The Born of Social Networks New types of Social Networks My Social Networks Research related

Private Social Network

• General social network may not allow you to upload some “sensitive” information or private information. However private social network allow you to create a group-centric SNS in which you can share more if you want.

Page 16: Social Network Something Interesting Ruizhi Gao. Contents The Born of Social Networks New types of Social Networks My Social Networks Research related

Sgrouples• Sgrouples (sounds like ‘scruples’) is our very own private, group-centric social

network designed to mimic how small groups of people interact in their real lives. Sgrouples allows you to easily post content to different groups based on your real life interests – friends, family, work, sports teams, and hobby groups.

Page 17: Social Network Something Interesting Ruizhi Gao. Contents The Born of Social Networks New types of Social Networks My Social Networks Research related

Reference

• http://blog.sgrouples.com/new-social-networking-sites/

• http://en.wikipedia.org/wiki/List_of_social_networking_websites

Page 18: Social Network Something Interesting Ruizhi Gao. Contents The Born of Social Networks New types of Social Networks My Social Networks Research related

Experience

Page 19: Social Network Something Interesting Ruizhi Gao. Contents The Born of Social Networks New types of Social Networks My Social Networks Research related

X-land Project

• We designed the Xland project as a 3D immersive blog. Xland was part of the CHIPS (CHina Innovation Program for Students) program sponsored by Sun Microsystems and the Chinese Education Department.

• It may be not called as a blog but a social space. Every user has its own room and we provide a open space like a community. People can decorate their own room and share information with each other. The most important thing is it’s not a client based but can be accessed from your web browser, which is very hard at that time. You need to consider very carefully about using the resource.

Page 20: Social Network Something Interesting Ruizhi Gao. Contents The Born of Social Networks New types of Social Networks My Social Networks Research related

Trailer

Page 21: Social Network Something Interesting Ruizhi Gao. Contents The Born of Social Networks New types of Social Networks My Social Networks Research related

Functions

Friend list and live chat HUD Albums

Keyboard piano

Decorate your room

Page 22: Social Network Something Interesting Ruizhi Gao. Contents The Born of Social Networks New types of Social Networks My Social Networks Research related

Functions

Background music of your room Dairy

Visitor

We tried to transfer the “my page” in facebook Into “my room” in a 3D world. Good idea but impractical like what Kaifu Li did 3D browser

Our blog:http://blogs.openwonderland.org/2010/07/23/xland-3d-blog/

Page 23: Social Network Something Interesting Ruizhi Gao. Contents The Born of Social Networks New types of Social Networks My Social Networks Research related

AlienSandal

• A real start up project SoLoMo based social network cooperated with students in UC Berkley

• AlienSandal is a social platform based on Google Map, which enable users to trace and customize Life Track and connect to people who have similar tracks.

• Trailer http://www.youtube.com/watch?v=ACYFEsodwrY

Page 24: Social Network Something Interesting Ruizhi Gao. Contents The Born of Social Networks New types of Social Networks My Social Networks Research related

Different Types of check-in Free draw in the map

Page 25: Social Network Something Interesting Ruizhi Gao. Contents The Born of Social Networks New types of Social Networks My Social Networks Research related

Difficulties

• What is the most important thing in SNS?• Why do you want to use a SNS application?• Why can SNS do?

• Answer from VC, “friends, money and …”How to make your SNS popular? RECOMMENDATION

Page 26: Social Network Something Interesting Ruizhi Gao. Contents The Born of Social Networks New types of Social Networks My Social Networks Research related

Clustering

Page 27: Social Network Something Interesting Ruizhi Gao. Contents The Born of Social Networks New types of Social Networks My Social Networks Research related

Why clustering is useful• Grouping users in social networks is an important process that improves

matching and recommendation activities in social networks. The data mining methods of clustering can be used in grouping the users in social networks [1]

• [1]S. Alsaleh, R. Nayak, Y. Xu, “Grouping People in Social Networks Using a Weighted Multi-Constraints Clustering Method,” in Proceedings of WCCI 2012, June,10-15. Brisbane, Australia.

[2] GAN, G., MA, C. & WU, J. (2007) Data clustering: theory, algorithms,and applications. ASASIAM Series on Statistics and Applied Probability, 20, 219-230.

[3] NAYAK, R. (2011) Utilizing past relations and user similarities in a social matching system. Advances in Knowledge Discovery and Data Mining, 99-110. 

Page 28: Social Network Something Interesting Ruizhi Gao. Contents The Born of Social Networks New types of Social Networks My Social Networks Research related

First Question• How many cluster???????

K-means, Fuzzy C means, K-medoids …

Rule of Thumb [1]

But this is not reliable….

[1]Kanti Mardia et al. (1979). Multivariate Analysis. Academic Press.

Page 29: Social Network Something Interesting Ruizhi Gao. Contents The Born of Social Networks New types of Social Networks My Social Networks Research related

Cluster Estimation 1

• Stephen L. Chiu, “Fuzzy Model Identification Based on Cluster Estimation,” Journal of Intelligent and Fuzzy Systems, Vol.2, pp 267-278, 1994.

• Suppose we have a collection of n data points {x1,x2,…,xn}, in our case. For each data points, we can assign a potential value.

and

ra is a positive constant. We have ra = 0.5 here (suggested by paper).

and ||xi – xj|| is the distance function between data xi and xj.

2|| ||

1

i j

nx x

ij

P e

2

4

r

Page 30: Social Network Something Interesting Ruizhi Gao. Contents The Born of Social Networks New types of Social Networks My Social Networks Research related

Kendall tau Distance

• Kendall tau distance

Example:

Two sample data

R1: 1,2,3,4,5

R2: 3,4,1,2,5

There are 4 disorder between R1 and R2

The distance will be

For R1 For R2 Count

1<2 3<4

1<3 3>1 *

1<4 3>2 *

1<5 3<5

2<3 4>1 *

2<4 4>2 *

2<5 4<5

3<4 1<2

3<5 1<5

4<5 2<5

41

5(5 1)2

Page 31: Social Network Something Interesting Ruizhi Gao. Contents The Born of Social Networks New types of Social Networks My Social Networks Research related

Euclidean Distance

Two sample data

R1: 1,2,3,4,5

R2: 3,4,1,2,5

• De = 2 2 2 2 2(1 3) (2 4) (3 1) (4 2) (5 5)

Page 32: Social Network Something Interesting Ruizhi Gao. Contents The Born of Social Networks New types of Social Networks My Social Networks Research related

Hamming Distance

Two sample data

R1: 1,2,3,4,5

R2: 3,4,1,2,5

• DH = 1+1+1+1+0

1 and 3 are different

2 and 4 are different

3 and 1 are different

4 and 2 are different

5 and 5 are different

Page 33: Social Network Something Interesting Ruizhi Gao. Contents The Born of Social Networks New types of Social Networks My Social Networks Research related

Cluster Estimation 1

• After we have potential value for each data point, we select the highest potential as the first cluster center. Let be the data point of the first cluster center and be its potential value. Then we will revise the potential of each data point xi by

• In which, we have:

and

*1x

*1P

* 21|| ||*

1ix x

i iP P P e

2

4

br

1.5b ar r

Page 34: Social Network Something Interesting Ruizhi Gao. Contents The Born of Social Networks New types of Social Networks My Social Networks Research related

Cluster Estimation 1

• Next we select the data point with the highest remaining potential as the second cluster center. We further reduce the potential of each data point according to their distance to the second cluster center. In general, after the kth cluster center has been obtained, we revise the potential of each data point by

Where is the location of the kth cluster center and is its potential value.

* 2|| ||* i kx xi i kP P P e

*kx *

kP

Page 35: Social Network Something Interesting Ruizhi Gao. Contents The Born of Social Networks New types of Social Networks My Social Networks Research related

Cluster Estimation 1

• Every time we got the , we need to decide whether we should select that as a new center , we have following process.

if Accept as a cluster center and continue.

else if

Reject and end the selecting process

else

Let dmin = [shortest of the distance between and all previously found cluster centers]

if

Accept as a cluster center and continue the whole process

else

Reject and set the potential at to 0.

Select the data point with the next highest potential as new and re-test

end if

end if

In which, we have and (suggested in paper)

*kP

* *1kP P

*kx

* *1kP P

*kx

*min

*1

1k

a

Pd

r P

*kx

*kx *

kx*kx

0.5 0.15

*kx

Page 36: Social Network Something Interesting Ruizhi Gao. Contents The Born of Social Networks New types of Social Networks My Social Networks Research related

Running Example 1

• Suppose we have the following rankings set (which may represents different pages you viewed. 1 is page 1, 2 is page 2)

R1 = {1,2,3,4,5,6,7},

R2 = {1,2,4,3,5,7,6},

R3 = {7,6,4,5,3,1,2},

R4 = {7,6,5,4,1,3,2}.

First we will assign potential value for each of them by

We have P for each of them

P1 = 1.9311 P2 = 1.9336

P3 = 1.7451 P4 = 1.7496

2|| ||

1

i j

nx x

ij

P e

Page 37: Social Network Something Interesting Ruizhi Gao. Contents The Born of Social Networks New types of Social Networks My Social Networks Research related

Running Example 1

• We choose the highest value P2 = 1.9336, that is, we choose R2 as our first center and

• After this, we need to revise the potential value for each ranking by

So we got:

P1 = 0.06006

P2 = 0.0

P3 = 1.69382

P4 = 1.57027

We will choose P3 then go on the process.

* 21|| ||*

1ix x

i iP P P e

*1 1.9336P

Page 38: Social Network Something Interesting Ruizhi Gao. Contents The Born of Social Networks New types of Social Networks My Social Networks Research related

Running Example 1

• Since we have and

P3 = 1.69382 which is greater than

Refer to if

Accept as a cluster center and continue.

So P3 will be our second center.

Then we will revise the potential value again by

Then we will have

P1 = -0.02683

P2 = -0.04499

P3 = 0.0

P4 = 0.085356

So P4 will be our next choice.

*1 0.966833P *

1 0.290049P *

1 0.966833P

*kx

* 22|| ||*

1ix x

i iP P P e

* *1kP P

Page 39: Social Network Something Interesting Ruizhi Gao. Contents The Born of Social Networks New types of Social Networks My Social Networks Research related

Running Example 1

• Since we have and

P4 = 0.085356 which is less than

Refer to if

Reject and end the selecting process

So we will stop the whole process

• Out of 4 data set, we have estimate 2 clusters and 2 centers which are R2 and R3, this result makes sense.

*1 0.966833P *

1 0.290049P

*kx

*1 0.290049P

* *1kP P

Page 40: Social Network Something Interesting Ruizhi Gao. Contents The Born of Social Networks New types of Social Networks My Social Networks Research related

• ZHOU Shi-bing, XU Zhen-yuan, TANG Xu-qing, “New method for determining optimal number of clusters in K-means clustering algorithm,” Computer Engineering and Applications, Vol.46, pp 27-31, 2010.

• Beliakov, Gleb and King, Matthew 2006, Density based fuzzy c-means clustering of non-convex patterns, European journal of operational research, vol. 173, no. 3, pp. 717-728.

• Suppose we have a collection of n data points {x1,x2,…,xn}, in our case, they are the ranks. For each two data points, we calculate the distance.(We use Kendall tau distance here)

Step 1(in paper):

Randomly choose one data point as the first center c1.

Step 2(in paper):

Choose the data point which has largest distance from the first center c1 as the second center c2.

Cluster Estimation 2: Max-min Distance

Page 41: Social Network Something Interesting Ruizhi Gao. Contents The Born of Social Networks New types of Social Networks My Social Networks Research related

• Step 3:

Get the distance between the rest of the data and all the centers.

then we get

Then, if then, we choose xi as our next center.

, we choose 0.6

Repeat this process until we cannot find any more centers.

Cluster Estimation 2: Max-min Distance

ij i jd x c 1,2.... (# )j k center

1,2.... (# )i n data1 2min( , ... )i i i ikd d d d

1 2max{ } || ||iD d c c

Page 42: Social Network Something Interesting Ruizhi Gao. Contents The Born of Social Networks New types of Social Networks My Social Networks Research related

Running Example 2

• Suppose we have the following rankings set.

R1 = {1,2,3,4,5,6,7},

R2 = {1,2,4,3,5,7,6},

R3 = {7,6,4,5,3,1,2},

R4 = {7,6,5,4,1,3,2}.

• The Kendall tau distance between each other is:

R1 R2 R3 R4

R1 0 0.095238 0.904762 0.904762

R2 0.095238 0 1 0.809524

R3 0.904762 1 0 0.190476

R4 0.904762 0.809524 0.190476 0

Page 43: Social Network Something Interesting Ruizhi Gao. Contents The Born of Social Networks New types of Social Networks My Social Networks Research related

Running Example 2

• The largest distance is between R2 and R3 which is 1.

So we choose these 2 as two centers then

Since

We stop the process and choose R2 and R3 as the centers.

1 1 2

4 1 2

0.095238

0.1904

min( , )

min( , 7) 6i i

i i

d d d

d d d

1 20.190476 0.max{ } || ||6iD d c c

Page 44: Social Network Something Interesting Ruizhi Gao. Contents The Born of Social Networks New types of Social Networks My Social Networks Research related

Pros and Cons

• Estimation 1ra, at least 3 parameters needs to be decided in the process, and all of this parameters are application sensitive. For now, there is no adaptive way to adjust such parameter.

• Estimation 2The problem is on the initial center selection. The result will be affected by the noise data point

, ,

Page 45: Social Network Something Interesting Ruizhi Gao. Contents The Born of Social Networks New types of Social Networks My Social Networks Research related

• Question