NoTube Project Collecting Data Social Web

Preview:

DESCRIPTION

NoTube Workpackage 3 status update: a quick survey on the NoTube approach to gathering user data from the Social Web and representing them as a RDF graphs.

Citation preview

The Beancounter: collecting data from the

Social Web

Davide Palmisano, Michele Minno and Michele Mostarda

3rd Project Meeting - 16/09/2009 @ Amsterdam

a ten-minutes long update on the WP3 status

a (very) short ToC

User profiling and context models

Where we are

user data gathering in the Social Web

the NoTube Beancounter: a general approach

a simple demonstration

Where we are going

Linked Music Explorer and the Beancounter

collecting data in the Social Web

User profiling and context models

User profiling and context models

collecting data in the Social Web

extremely high heterogeneity:

User profiling and context models

collecting data in the Social Web

extremely high heterogeneity:

different data models

User profiling and context models

collecting data in the Social Web

extremely high heterogeneity:

different data models

syndications

extremely high heterogeneity:

different data models

syndications

auth technologies

User profiling and context models

collecting data in the Social Web

a possible dev process:

choose a “social” application:

User profiling and context models

the Beancounter approach

choose a “social” application:

write code to:

User profiling and context models

the Beancounter approach

a possible dev process:

choose a “social” application:

write code to:implement the auth policy

User profiling and context models

the Beancounter approach

a possible dev process:

choose a “social” application:

write code to:

parse the responseimplement the auth policy

User profiling and context models

the Beancounter approach

a possible dev process:

choose a “social” application:

write code to:

translate it in RDF and store it

repeat for all the stuff in the Social Web

parse the responseimplement the auth policy

User profiling and context models

the Beancounter approach

a possible dev process:

choose a “social” application:

write code to:

translate it in RDF and store it

repeat for all the stuff in the Social Web

parse the responseimplement the auth policy

User profiling and context models

the Beancounter approach

a possible dev process:

choose a “social” application:

write code to:

translate it in RDF and store it

repeat for all the stuff in the Social Web

parse the responseimplement the auth policy

User profiling and context models

the Beancounter approach

a possible dev process:

choose a “social” application:

write code to:

translate it in RDF and store it

repeat for all the stuff in the Social Web

parse the responseimplement the auth policy

User profiling and context models

the Beancounter approach

a possible dev process:

choose a “social” application:

write code to:

translate it in RDF and store it

repeat for all the stuff in the Social Web

parse the responseimplement the auth policy

User profiling and context models

the Beancounter approach

a possible dev process:

choose a “social” application:

write code to:

translate it in RDF and store it

repeat for all the stuff in the Social Web

parse the responseimplement the auth policy

User profiling and context models

the Beancounter approach

a possible dev process:

a bit boring, isn’t it?

instead, what I really want is:

a framework that allows me to reduce at minimum the development effort

a general architecture that embraces the heterogeneity

allowing a decoupled and third party development

User profiling and context models

the Beancounter approach

User profiling and context models

the Beancounter approach

User profiling and context models

The NoTube Beancounter principles:

an engine that allows to extract and aggregate users social data

representing the data with RDF and storing them in a preferred triple store

fully accessible with a set of REST APIs

a general architecture with hot-pluggable components (tubelets and modelets)

User profiling and context models

the Beancounter architecture

User profiling and context models

the Beancounter architecture

User profiling and context models

the Beancounter architecture

User profiling and context models

the Beancounter architecture

User profiling and context models

the Beancounter architecture

User profiling and context models

the Beancounter architecture

User profiling and context models

the Beancounter architecture

a quick demo around the following scenario:

an instance of the Beancounter is running

an administrator wrote a Tubelet for BrightKite and want to upload it to the Beancounter

Davide wants to let the Beancounter storing his data from his Brightkite account

User profiling and context models

What you are going to see

How will Linked Music Explorer interact with an instance of the Beancounter?

User profiling and context models

Beancounter interactions

How will Linked Music Explorer interact with an instance of the Beancounter?

User profiling and context models

Beancounter interactions

How will Linked Music Explorer interact with an instance of the Beancounter?

User profiling and context models

Beancounter interactions

User profiling and context models

Further details

architecture

how the Beanconter interacts with other components?

recommendation

how to use the “beans” to provide content recommendation?

what kind of APIs?

backup

Architecture sketch