G&N Unit 8

If you had to organize a viral marketing campaign in a social network, which action would be more effective?

• Start the diffusion at the nodes with highest centrality.

• Start the diffusion at randomly selected nodes.

• Localize the maximal cliques, and start the diffusion at all their nodes.

• Start the diffusion at the nodes with lowest threshold of exposure.

The figure represents an intermediate step of a diffusion process (threshold of exposure at nodes, intensity of influence at edges). Which nodes will become adopters at next step?

• node 5

• node 7

• Both, nodes 5 and 7

• No node will adopt the innovation

Which of the following sentences is TRUE?

• The adoption speed is always constant.

• The size of adoption and adoption speed are smaller in dense networks than in sparse networks.

• The size of adoption and adoption speed are bigger in dense networks than in sparse networks.

• The addoption speed decreases when the diffusion enters in a a highly connected cluster.

In a SIR model, at time t we have the following observations:

#infected is 50.

#susceptibles is 200.

#recovered is 10.

The daily interaction rate is 2.

The probability of contagion is 10%.

The recovery rate is 100% (no mortality).

The average duration of the disease is 10 days.

Question: How many new recovered will exist at time t+1?

• 2

• 10

• 100

• 5

Given these plots of total infected people at time t, which one corresponds to a SI (Susceptible-Infected) model

• 
• 
• 
• All of them illustrate a SI model

• 
• 
• 
• There is no effect

Which immunization strategies would be effective?

• Finding the potential main spreaders (nodes with high centrality) and vaccinate them.

• Finding communities and isolate them (remove the edges inter-community)

• Detecting high interactions (edges with high centrality) and isolate them.

• All the actions are good.

In message passing ...

• The nodes are updated depending on the interaction coefficient.

• The aggregate function combines the messages of the neighbouring nodes and the update function combines the attributes in the previous iteration with the result of the aggregate.

• The new values of the nodes are obtained as the ratio between the aggregate and the update function.

• The update of the nodes follows a SIR model.