streams and filtering

71
Streams and Filtering [email protected]

Upload: sangwon-yang

Post on 11-May-2015

1.853 views

Category:

Technology


2 download

DESCRIPTION

Semantic Streams and Social Filtering

TRANSCRIPT

Page 1: Streams and filtering

Streams and Filtering

[email protected]

Page 2: Streams and filtering

i, my, me, mine

•Sangwon Yang

- //sangwon, @sangwon ...

- was a service designer @ NHN, RTWK, Nurien

- am a Master Student @ BiKE Lab.

Page 3: Streams and filtering

Today, My talk is...

•Social Web

•Semantic Streams

•Filtering

Page 4: Streams and filtering

Social Web

Page 5: Streams and filtering

http://www.flickr.com/photos/briansolis/2735401175/sizes/l/

Page 6: Streams and filtering

Social Web

• The Social Web is currently used to describe how people socialize or interact with each other through the World Wide Web. Such people brought together through a variety of shared interests.

- wikipedia, social webhttp://en.wikipedia.org/wiki/Social_web

Page 7: Streams and filtering

Social Web

•The Social Web consists of people who are connected by a shared object.

- Social Graph : the network of connections and relationships between people.

- Social Object : the reason, people are talking to each other.

Page 8: Streams and filtering
Page 9: Streams and filtering

Internet Trends, April 12, 2010, Morgan Stanley

Page 10: Streams and filtering

Current Social Web

Identity

Friends

Streams What's going on

Who I know

Who I am

Page 11: Streams and filtering

identity

friendsstreams

Page 12: Streams and filtering

Simple & Powerful Paradigm

•Archiving / Displaying Your social gesture

•Browsing / Navigating Your Friends social gesture

Page 13: Streams and filtering

Streams

•The cleanest and simplest way to navigate a Social Web

•Streams are rapidly changing sequences of information around the topic

- microblogs, feeds, even data streams.

•Streams are real-time, flowing, dynamic activity of information

Page 14: Streams and filtering

Twitter

•via @channyun #chirp

- users 105m

- 0.3m subscribe per day

- 55m tweets per day

- 600m queries per day

Page 15: Streams and filtering

Facebook

•Facts & Figures

- http://www.allfacebook.com/2010/03/facebook-summarized-in-a-single-picture/

Page 16: Streams and filtering

NHN

NHN Facts, 03, 2010

Page 17: Streams and filtering

Social Web Issues

Page 18: Streams and filtering

Current Social Web

•While current social web gained a lot of interest of the Web. It also raises various Issues.

- multitude of online identities and social graphs

- Isolated data silos.

- Unstructured contents.

- Scattered conversations.

Page 19: Streams and filtering

We need a way...

•for representing and navigating the content items in these sites. Also to better enable user to access to multiple sites and to content-creation facilities on the Web, interoperability among social web is required in term of both the content objects and the person-to-person networks expressed on each site.

•This requires representation mechanisms to interconnect people and objects on the Web in an interoperable and extensible way.

Page 20: Streams and filtering

Semantic Streams

Page 21: Streams and filtering

Semantic Streams

•It allows users to merge user's multitude of online identities, social graphs and interconnect from isolated data silos to structured contents

•It enables generation of streamed RDF data from existing Social Web(streams) sites.

Page 22: Streams and filtering

Identity and Social Graph

Page 23: Streams and filtering

Identity

•Social Web users can have multitude online accounts, passwords and profiles

- maintaining a multitude of profiles is cumbersom and consuming

- each platform contains a different facet of their online identity

Page 24: Streams and filtering

Distributed Identities

twitter:sangwon fb:sangwon.yang

me2day:olli google:sangwon.yang

Page 25: Streams and filtering

Distributed Social Graphs

Page 26: Streams and filtering

FOAF(Friend of a Friend)

- The FOAF is creating a Web of machine-readable pages describing people, the links between them and the things they create and do

- It allow for a more expressive description of a social network, enabling the use of heterogeneous nodes and links denoting different types of objects and different types of relationships.

Page 27: Streams and filtering

FOAF SPEC

FOAF terms as illustrated by Dan Brickley

Page 28: Streams and filtering

Describing People

•Basic Information

- name, sex, birthday, nickname...

•Personal Information

- interest, organization, weblog...

•Contact Information

- online account, email, IM, phone...

Page 29: Streams and filtering

Merging Identities

•foaf:account

•owl:sameAs

Page 30: Streams and filtering

foaf:account

Page 31: Streams and filtering

foaf:account

twitter:sangwon

fb:sangwon.yang

flickr:29799686@N04

me2day:olli

google:sangwon.yang

Sangwon Yang

foaf:account

foaf:account

foaf:account

foaf:account

foaf:account

Page 32: Streams and filtering

owl:sameAs

:me owl:sameAs twitter:sangwon

owl:sameAs fb:sangwon.yang

owl:sameAs me2day:olli

owl:sameAs google:sangwon.yang

Sangwon Yang

twitter:sangwonfb:sangwon.yang

me2day:olligoogle:sangwon.yang

Page 33: Streams and filtering

Describing Relationships

•foaf:knows is the most important property

- used to create a friend link

•foaf:knows, (extended with the Relationships vocabulary

- :me foaf:knows :you

- :me rel:friendOf :you

Page 34: Streams and filtering

http://example.org/foaf.rdf#Jaeho

!"#"!$%&'

http://example.com/foaf.rdf#Junho

!"#"!$%&'

http://xmlns.com/foaf/0.1/knows

!"#"!$%&'

Page 35: Streams and filtering

Sangwon Yang

twitter:sangwon

twitter:echo4ngel

twitter:soo1005

twitter:egero

twitter:brotherpark

twitter:johnbreslin

foaf:knows

foaf:knows

foaf:knows

foaf:knows

foaf:knows

foaf:knows

Page 36: Streams and filtering

Distributed Social Graph

Sangwon Yang

twitter:sangwon

twitter:echo4ngel

twitter:soo1005

twitter:egero

twitter:brotherpark

twitter:johnbreslin

foaf:knows

foaf:knows

foaf:knows

foaf:knows

foaf:knows

fb:701611529

fb:1254820102

fb:1238686601

fb:1529605506

fb:1255152455

fb:618865181

fb:1254708141

foaf:knows

foaf:knows

foaf:knows

foaf:knows

foaf:knows

foaf:knows

me2:olli

me2:ymjkim

me2:purumae

me2:neddang36

foaf:knows

foaf:knows

foaf:knows

Page 37: Streams and filtering
Page 38: Streams and filtering

Interlinking Social Graphs

twitter:sangwon

twitter:echo4ngel

twitter:soo1005

twitter:brother

foaf:knows

foaf:knows

foaf:knows

fb:sangwon.yang

fb:jaeho.yang

fb:junho.yangfoaf:knows

foaf:knows

me2:olli

me2:echo4ngel

me2:egero

foaf:knowsfoaf:knows

Sangwon Yang

owl:sameAs owl:sameAs

owl:sameAs

Page 39: Streams and filtering

Streams

Page 40: Streams and filtering

Streams

Page 41: Streams and filtering

Streams

•Most Social Web services act as closed worlds like.

•At present, the content of Streams does not carry any semantics, making its use difficult

Page 42: Streams and filtering

SIOC(Semantically-Interlinked Online Communities)

•SIOC aims to enable the integraion of online community information

•SIOC fully describe the contents and structure of community sites

Page 43: Streams and filtering

The main classes and properties in SIOC

SIOC

Page 44: Streams and filtering

Streams

Page 45: Streams and filtering

Structured Streams

http://twitter.com/sangwon/status/10161248992

sioct:MicroBlogPost http://twitter.com/sangwonMon, 08 May 07:42:05 +0000 2010

RT @SNUBIKE This week's BiKE Talk was given by

@channyun about an ontology based bug analysis http://is.gd/9W3DB

sioc:UserAccount 1476571 Datatype: xsd:integer sangwonhttp://a3.twimg.com/

profile_background_images/87445587/me2photo.jpg

Datatype: xsd:anyURI

Language: enrdf:type dc:created sioc:has_creator

sioc:content

rdf:type sioc:id rdfs:labelsioc:avatar

Page 46: Streams and filtering

Structured Streams

Page 47: Streams and filtering

Yoonyoung Nam

Sangwon Yang

foaf:Person

foaf:knows rel:collaboratesWith

twitter

sangwon

sioc:UserAccount

foaf:Document

foaf:accountServiceHomepage

foaf:holdsAccount

sangwon on twitter

sioc:Forum/sioct:Microblog

!"!"1

!"!"3sioc:creator_of

sioc:creator_of

sioc:owner_of

sioc:Post

sioc:Post sioc:container_of

sioc:container_of

foaf:Person

twitter

yoonyoung

sioc:UserAccount

foaf:Document

foaf:accountServiceHomepage

foaf:holdsAccount

yoonyoung on twitter

sioc:Forum/sioct:Microblog

!"!"2

sioc:Post sioc:container_of

sioc:creator_of

sioc:owner_of

sioc:reply_ofsio:reply

Page 48: Streams and filtering

Filtering

Page 49: Streams and filtering

Filtering

•We can follow many streams.

•Cramming all our separate stream into merge activity stream get too noisy and overwhelming to use.

Page 50: Streams and filtering

Streams

socialfeed

CBCB CB

CB

CBCB

CB CB CB

CB

CB

CB

CBCB

CB

CB

CB

CB

CB

CB

CB

CB

CBCB

CB

CB

CBCB

CB

CB

CB

CBCB

conversations

conversations

conversations

conversations

Page 51: Streams and filtering

Filtering

•We need away that can adapt to filter the Stream differently based on what we now care about.

Page 52: Streams and filtering

Social Filtering

•We need ways that can adapt to filter the Stream differently based on what we now care about.

- to see just the things that are really fun, important, popular or likely to be of interest and value to users, instead of everything

Page 53: Streams and filtering

Social Filtering

Page 54: Streams and filtering

Approach

•Social Filitering

- filters information based on User Affinity, which is measured by degree of user relationship, and frequency of communications or interactions between users

- filters information based on User Topics(or user context)

Page 55: Streams and filtering

User Affinity•Measuring degree of user relationship

- based on user affinity among multiple Social Graphs

- Explicit

‣ based on merged social graphs

- Implicit

‣ based on user interactions : frequency of communications and methods of communication

Page 56: Streams and filtering

Merged Social Graphs

Page 57: Streams and filtering

degree of user relationship

Sangwon Yangtwitter:sangwon

fb:701611529flckr:29799686@N04

me2:olli

Yeonju Ahntwitter:dongmulwon

fb:1529605506

Doeyull Kimtwitter:purumaefb:1238686601me2:purumae

HK Kimtwitter:egero

flckr:11095462@N04

Youngmin Kimtwitter:ymjkimfb:618856181me2:ymjkim

Daegon Kimfb:1254820102

3

3

12

2

Page 58: Streams and filtering

User Interactions

•Frequency of Communications

- measuring of frequency of conversations between users

•Methods of Communication

- following, favorite, RT, Reply, DM

Page 59: Streams and filtering

twitter:yoonyoungtwitter:sangwon

follow

Contents

publish

reply : 2rt : 1

favorite : 1

Measuring of user Interactions

Page 60: Streams and filtering

User Topics

•What is user talking about?

- Keywords, tag, link, metadata...

•Social Context

- SC = f(P,E) P:Person/Personality E:Environment

Page 61: Streams and filtering
Page 62: Streams and filtering

Social Filtering

•Who

- user(actor)

•With Whom

- based on user social graphs

•What

- topics

- user context

Page 63: Streams and filtering
Page 64: Streams and filtering
Page 65: Streams and filtering

Yoonyoung Nam

Sangwon Yang

foaf:Person

foaf:knows rel:collaboratesWith

twitter

sangwon

sioc:UserAccount

foaf:Document

foaf:accountServiceHomepage

foaf:holdsAccount

sangwon on twitter

sioc:Forum/sioct:Microblog

!"!"1

!"!"3sioc:creator_of

sioc:creator_of

sioc:owner_of

sioc:Post

sioc:Post sioc:container_of

sioc:container_of

foaf:Person

twitter

yoonyoung

sioc:UserAccount

foaf:Document

foaf:accountServiceHomepage

foaf:holdsAccount

yoonyoung on twitter

sioc:Forum/sioct:Microblog

!"!"2

sioc:Post sioc:container_of

sioc:creator_of

sioc:owner_of

sioc:reply_ofsio:reply

Concept

Concept

sioc:topic

sioc:topic

skos:narrower

Page 66: Streams and filtering

Components of Semantic Streams

Page 67: Streams and filtering

Layers of Semantic Streams

•Unified User Identity and Social Graph

•Stream Parsing Engine

•Smart Data

•Republishing Activity Streams

•Conversations

•Social Filtering

Page 68: Streams and filtering

Streams Parsing Engine

•Import Streams

- the process importing streams and parsing it

- the process analyzing streams

•Discovery

- the process in which machines find out information about the Web resources

- metadata crawling

Page 69: Streams and filtering

Smart Data

•Smart data is a key factor for Social Filtering

- User implicit/explicit data

- conversations data

- streams & metadata

- context data

Page 70: Streams and filtering

Blocks of Semantic Streams

Page 71: Streams and filtering

생유!