1. Social scope: Enabling Information Discovery On Social Content Sites
Presenters: Sahar Delroshan
Rezan Amiri
Azad university of Kurdistan

2. outline Overview Of Social Scope Case Study With Yahoo! Travel
Motivating Example
Social Scope Platform
Layers of social Scope
Data and query
Model of Social Content Management
Conclusion
Azad university of Kurdistan

3. Overview Of Social Scope
Recently, many content sites have started encouraging their users to engage in social activities such as adding buddies on Yahoo! Travel and sharing articles with their friends on New York Times.This has led to the emergence of social content sites,which is being facilitated by initiatives like Open ID and Open Social.
Azad university of Kurdistan

4. Content Site Are Popular
Azad university of Kurdistan

5. Social Sites Are Popular!
Azad university of Kurdistan

6. Social Content Sites
Azad university of Kurdistan

7. Case Study With Yahoo ! Travel
Y!Travel is a typical content site that is gradually evolving into a social content site.
Y! travel Data:
Y ! Travel maintains a comprehensive set of travel objectes : cities , restaurants, etc.
Y! Travel Queries:
Users interact with Y!Travel through a search interface, where they enter a set of keywords and obtain a list of travel objects considered relevant to their queries
Azad university of Kurdistan

8. Case Study With Yahoo ! Travel
we detect location terms in queries and classify each query into three classes:
general, categorical, and specific.4 General queries are those containing
terms like “things to do”, “attraction”, or just a location by
itself. Over 50% of the queries fall into this class, and about 60% of
those queries contain a location. Categorical queries refer to those
containing terms like “hotel”, “family”, “historic”, etc. About 30%
of the queries fall into this class and a majority of them mention a
location. Finally, there are also about 8% of the queries looking for
specific destinations like “Disneyland” and “Yosemite Park”.
Azad university of Kurdistan

9. Motivation Example Example : Two major paradigms :
Search for “Denver attractions”
Two major paradigms :
1-semantic relevance
2- social relevance
EXAMPLE 1. John is in Denver for a conference. Having one
day free, he visits Y!Travel and searches for “Denver attractions”.
John has in the past visited quite a few baseball fields on
Y!Travel and has many friends on Facebook with interests in
“baseball”. With this knowledge, Y!Travel recommends to him
“B’s Ballpark Museum” (a small baseball museum in the suburb),
“Coors Field” (home field of the Rockies), as well as the upcoming
baseball game “Yankees vs Rockies” to be played at Coors Field,
which is fetched from Y!Sports.
Example 1 represents one out of three queries on Y!Travel.
However, the traditional information retrieval approach fails for it
because there are often many objects that are semantically relevant
to John’s query and no ranking mechanism (e.g., tf-idf measure)
based on pure semantic relevance can differentiate them. It is
therefore imperative for the system to incorporate social relevance,
which considers John’s social profile and connections, to decide
which attractions he will prefer. Essentially, information discovery
on social content sites requires the integration of two major
paradigms: semantic relevance with respect to a query and social
relevance in the spirit of recommendations. The former scopes the
discovery to information relevant to John’s current needs as expressed
by him, while the latter identifies the information most appealing
to John as a user. Indeed, “B’s Ballpark Museum” may not
be a major attraction, yet John, being a baseball fan, is likely to
enjoy a visit to it. Example 1 further illustrates another important
desideratum: the need to retrieve relevant information from external
social or content sites that are physically and administratively
separate from Y!Travel, which is becoming possible because of
various initiatives like OpenSocial
Azad university of Kurdistan

10. Insights from the Examples
Integration of three major paradigms for discovery social content sites:
1. Keyword search
2. Database-style querying
3. Recommendations
Personalized search
Faceted search
Azad university of Kurdistan

11. Information organizer Information discoverer
Social Scope Platform
Information Presentation
result selector
Information organizer
User Interface
Information Discovery
Query /
Result
Social Content Admin
Content analyzer
Information discoverer
User
Activity
Manager
Data Manager
Content Management
Social
Content
Graph
Activities
Content Integrator
OpenSocial API
OpenSocial API
Facebook
Y! IM
Y! Sports
Azad university of Kurdistan

12. Architectural Vision The core is the social content graph
users, objects, and various connections among them
Information in the graph
locally owned ( destinations in Y! Travel)
Externally integrated ( friendship connection obtained from Facebook )
derived ( links describing similarities between users)
در هسته اصلی آن گراف محتوایی اجتماعی وجود دارد که نشان دهنده کاربران و موضوعات و ارتباطات متنوع بین آنها است
اطلاعات موجود در گراف شامل :
1- محلی است ( مانند مقصد ها)
2- ارتباطات یکپارچه خارجی ( رابطه دوستی در فیس بوک)
3- اطلاعات مشتق شده ( لینک های که شباهت بین کاربران را توصیف می کند)
Azad university of Kurdistan

13. Three layers of SocialScope system
Content Management
Information Discovery
major goal: efficient and flexible mechanism
logical algebra
Each operator in the algebra takes social content graphs as input and outputs a social content graph
Information Presentation
هدف اصلی لایه دو (کشف اطلاعات) :
ایجاد یک راهکار منعطف و کاربردی برای ساختن یک گراف محتوایی اجتماعی است
Azad university of Kurdistan

14. Content Management Content Integrator Data Manager Activity Manager
facilitate the incorporation of social info from remote sites
Data Manager
maintenance and retrieval of the social content graph
Activity Manager
when and how refresh parts of the social graph
Categorizing users based on their activities
دو وظیفه اصلی دارد :
1- تلفیق اطلاعات اجتماعی را با سایتهای ی از راه دور از طریق بخش اول آسان می کند
2- نگهداری و بازیابی گرافهای محتوای اجتماعی از طریق بخش دوم انجام می شود
علاوه بر این نیز
از طریق بخش سوم به وسیله دسته بندی کردن کاربران بر اساس فعالیتهایشان کمک می کند که تصمیم بگیرند کی و چگونه فسمتهایی از گراف را تازه و بروز کنند
Azad university of Kurdistan

15. Content Management three major categories of data: site content
users’ social profiles and connections
users’ site-specific social activities
-site content is the content that users are interested in when they visit the
social content site
-Social profiles
and connections are the information regarding the users themselves
(e.g., name, education, etc.) and their explicit social connections
(e.g., friends, classmates, colleagues, etc.)
-site-specific social activities are the activities users perform on the site content. For example, in Y! Travel, users visit and browse destinations
How to effectively and efficiently manage the three categories of data is at the heart of challenges to be addressed by the Content Management layer of our SocialScope system
Azad university of Kurdistan

16. Information Discovery
Content Analyzer
derives new nodes and links through various analyses of the raw social content graph
automatically by the system itself
Social Content Administrator
Information Discoverer
Meaningful Social Graph (MSG)
این بخش تحلیل گر محتوا :
گره های جدید و لینک ها را از طریق تحلیل های گوناگونی که بر روی گراف های محتوایی اجتماعی انجام می دهد مشتق می شود که این تحلیل یا به صورت اتوماتیک از طریق خود سیستم انجام می گیرد یا از طرف مدیر محتوای اجتماعی
بخش دوم :
پرس و جو های کاربران را تجزیه می کند و یک نمایش داخلی را برای آنها شکل می دهد و آنها را بر روی گراف محتوای اجتماعی ارزیابی می کند که نتیجه تمام این کارها به یک گراف پر معنی محتوای اجتماعی تبدیل می شود
Azad university of Kurdistan

17. Information Presentation
MSG as input
dynamically organizes the results
Information Organizer and Result Selector
grouping
Social Grouping ,Topical Grouping , Structural Grouping
Ranking
identifies appropriate mechanisms for ranking and selecting results
به طور پویا نتایجی را سازماندهی می کند که کاربران بتوانند اکتشافات موثری داشته باشند
موضوعات مناسبی ( ساختاری یا اجتماعی ) را برای نتا یج گروپینگ آماده می کند
مکانیزمهای مناسبی برای انتخاب کردن نتایجی که از گروهای مختلف بدست می آید را شناسایی می کند
زمانی هم که چندین گروههای ارائه وجود دارند بخش سازماندهی اطلاعات تصمیم می گیرد که کدام گروه بیشتر به کاربران مربوط است و به آن اطلاعات نیاز دارد
Azad university of Kurdistan

18. Information Presentation
The right presentation can help a user
explore the information more effectively, especially when she is not sure about exactly what she wants
Our vision is to build :
dynamic result exploration framework
recommender systems:
presentation is an important aspect
has direct implications on building users’ trust
-In search, presentation is primarily in the form of a single ranked
list of results, where a result’s rank reflects its degree of relevance
to the input query
-In recommender systems, presentation is an important
aspect and has direct implications on building users’ trust
and giving them incentives to participate in more activities
-The rich activities and user connections provide lots of opportunities for generating recommendations
Azad university of Kurdistan

19. Data And Query Model We model an instance of a social content
site as a social content graph
Consists of nodes and links
Connections and activities
physical and abstract entities
User
Topics
Friendship and tagging action
Azad university of Kurdistan

20. Data And Query Model flexible (schema-less) typing system
the type attribute have multiple values
n1 = {id=1; type=‘user, traveler’; name=‘John’}
catalog of basic types
user, item, topic, group for nodes
connect (e.g., friend), act (e.g., tag, review, click,etc.), match, belong for links
n2 = {id=2; type=‘item, city’; name=‘Denver’; keywords=‘skiing’}
are two nodes representing our traveler John and the city Denver,
respectively, in Example 1.
Similarly, l12(n1, n2)
= {id=12;type=‘act, tag’; date=‘ ’; tags=‘rockies baseball’} is a link recording the activity that John tagged Denver with tags ‘rockies baseball
Azad university of Kurdistan

21. Data And Query Model Social Content Graph overlay of sub-graphs
activity graph
network graph
topical graph
Queries
Users interact with SocialScope by specifying a query on content and structure
the activity graph, which maintains users’ activities on items
the network graph, which maintains social connection
the topical graph, which maintains links from users or items to derived
semantic groups or topics.
the term structure refers to the attribute/value
pairs associated with nodes and links.
Azad university of Kurdistan

22. Logical Algebra Basic operators
Node Selection (σN), Link Selection (σL)
Composition, Semi-Join
Node Aggregation, Link Aggregation
Example search
- Each operator in the algebra takes
social content graphs as input and outputs a social content graph
- Both operators Node Selection (σN), Link Selection (σL)
take a condition C and an optional scoring function S as
parameters, and a (social content) graph as input
- The condition C
consists of a list of structural conditions (e.g., {type=‘city’, rating‘0.5’}) and a set of keywords (e.g., ‘Denver attraction’)
Both Composition and Semi-Join “connect” links in their input graphs. However, composition generates new links while semi-join simply filters away unwanted links.
-One important feature of aggregation is the creation of new information (aggregation results) that need to be stored and maintained.
-the aggregation function takes
as input a collection of links and produces as output a value to be associated with the attribute a t t
-a t t is the destination attribute
-two classes of aggregations
SAF map a set of links to a set of scalars
NAF map a set of links to a numerical scalar value
-The definition of the Link Aggregation operator is analogous to Node Aggregation
except for two major differences:
First, link aggregation changes the structure of an input graph: it replaces a set of links between a given src and tgt node by a new link.
Secondly, the result of the aggregate computation is assigned as a destination attribute of the newly created link.
Azad university of Kurdistan

23. A Simple Search Task John’s friends People visited Denver
who visited Denver
Their activities
Azad university of Kurdistan

24. Model for Social Content Management
-Decentralized Model: In this model, each social content site
maintains their own social information, including storing the user
profiles and social connections, and effectively manages the entire
social content graph internally
-social graph decentralization means it is necessary
for users to establish their social connections multiple times
on many different sites, even though most of those connections are
the same.
-Decentralization establishing a social graph with critical mass is incredibly difficult
-Closed Cartel Model In this model, users establish and maintain their social profiles and connections at a few of the dominant social sites and let
those sites or third-party applications, which are developed specifically
for those sites, fulfill their content needs. Facebook is the
prime example of this model.
-There are two major implications for users:
First, users no longer have to maintain
many social profiles and establish the same social connections at
many different sites, which is a significant improvement over the
Decentralized Model.
Second, however, they are forced to have a
central online social presence, without which they won’t even have
access to the contents otherwise would have been available on content
sites.
-Decentralized Model is being replaced by the Cartel models
-which Cartel model will eventually come out on top
-small niche content sites (e.g., your neighborhood reading group)
will prefer the Closed Cartel model for ease of management
-While larger content sites (e.g., New York Times and Y!Travel) would
prefer the Open Cartel model
Azad university of Kurdistan

25. Conclusion
social content sites will increasingly become a part of the our online life
we proposed SocialScope, a logical architecture
three layers
an algebraic framework
three main categories of data
ways of presenting information help user understand the large variety of content discovered from social content sites
Azad university of Kurdistan

26. Strong point Logical architecture
Information discovery over social content site
Algebraic framework
Azad university of Kurdistan

27. weakness
Prototype and needs work
Azad university of Kurdistan

28. Questions?
Azad university of Kurdistan