Community detection in semantic graph

Community detection in semantic graph

who am I?

I defined my self as

?

attributeattribute

attribute

I can be define with my main attribute

attributeattribute

attribute

but more precisely with a melting pot of my attributes

attributeattribute

attribute

but I'm also defined by my social network

?

I can be defined by my main collaborators

but also by a melting pot of these collaborators

who are we?

instance of connected individuality

the majority generally define the group

but every collaboration build the identity of a community

different levels of cohesion

algorithme LP/RAK

don't we have better descriptors than random labels ?

I tag, so I am!

we tag and collaborate, who are we?

we are 'semantic web', 'web2.0', 'strategic watch'

propagation de tags"interaction creates similarity, while similarity creates interaction"

[mika 2005]éoliennes

éoliennes

éoliennes

éoliennes

but we still have

tag propagation

to infer this we need some knowledge

+ 1 = 32

+ 1 = 32

+ 1 = 6+ 23

algorithm LP/RAK

semantic propagation of tags

salt, water

pepper, winemustard

rugby, foot

foot, cine

hockey sport sport

sport

condiment

condimentcondiment

Applied to ademe's Ph.D. network

biomass

air pollution

soil pollution

energy efficiency

climat change