1.2 Definitions and Examples of Networks

Definitions and Examples of Networks
In order to study networks, it will be useful that we are

precise about what we mean.
Mathematically, a network is an object called a graph. It

consists of nodes, which are also called vertices. And there
are edges that connect the nodes. For example, in a social
network, each node is a person and an edge between two
people represents their friendship.
So let's discuss some more examples of networks. We have already mentioned the
World Wide Web, where each node is a page and the edges correspond to hyperlinks.
Similarly, the internet is a network. Here computers are the nodes and the edges are
network protocol interactions.
Another example of a physical network is the road network.

Traffic flowing through the network is a network process.
1
Also, most biological processes occur on networks. For example,
a neural network consists of neurons that are connected by
synapses. Neural activity, specifically, the pattern of electrical
signals or spikes is a network process.
Another example is a gene regulatory network. Here the vertices

are the genes and the edges represent regulatory effects.
In a food web, the nodes are species and the edges are drawn
between predator and prey.
In the recommendation system, relationships between items or

users can be represented by networks.
Finally, another example is the citation network with each node
representing a research article and edges representing citations.
So after seeing all these examples of networks, it's useful to try to categorize them.
The World Wide Web, for instance, is different from the internet in a basic way.
The World Wide Web is a directed network, whereas the internet is an undirected
network. Edges in directed networks have a direction. A website can link to another
2
website that does not link back. An edge on the internet represents a communication
channel. This is typically bi-directional. So we use an undirected edge in this case.
Food webs are directed networks given by the predator-prey relationship and similarly
for citation networks, gene regulatory networks.
We don't see many gazelles eating lions, so directed edge sounds
accurate.
On the other hand, the internet, power grids, neural networks, road networks, and
metabolic networks are mostly undirected. Although with that said, if there's a
one-way street, you would represent that with a directed edge.
In Facebook friendship is an undirected relationship. So you'd use an undirected graph.
Twitter, in contrast, allows you to follow people without them necessarily following
you. So it is best represented by a directed graph.
3
Even with an undirected network, there is a distinction between the type of network
representing the internet or power grid and neural networks or road networks.
⇒ While neural networks can have multiple connections between any two neurons,
power grids have at most one edge between any two nodes and they don't have any
self-loops. Such networks are known as simple networks.
Directed networks can also be categorized further based on whether or not they have
cycles.
⇒Networks that don't have any cycles in the underlying graph are called trees.
Examples are phylogenetic trees.
⇒If a directed network has no directed cycles, then it is called a directed acyclic graph
or DAG.
Notice that one can have a cycle in the graph, just not a direct cycle.
4
For example, food webs are usually DAGs since there is
an underlying hierarchy.
Smaller animals usually don't eat larger animals.
Also, citation networks are usually DAGs, since a paper

typically only cites earlier papers.
The information in a recommendation system can be

represented by a network with a special structure known
as the bipartite graph. In Netflix, we would have users on
one side and movies on the other side. Each user would
have an edge to the movies that they have rated. The
graph is called bipartite if the nodes can be partitioned
into two sets so that there are only edges between the
two sets and not within a set.
5
⇒So we have now introduced some of the most important classes of networks and
given many examples of networks we encounter on a daily basis.
We should also mention the different kinds of data collected on the network. For
example, in the World Wide Web, you can collect data on which page was accessed
from which other page. This data tells us about the presence of an edge. Alternatively,
you can collect data just on the nodes. For example, something like “How often was a
particular website visited?” This also contains information about the underlying
network.
In summary, below are some of the questions that we might be interested in when
studying a network.
⇒What would you like to look up on a social network?
For example, we might like to be able to identify people like Justin Bieber in order to try
to spread our tweets as efficiently as possible.
So one question we will discuss in this module is how to identify important nodes or
edges in the network. And what makes a node or an edge influential.
More generally, we will talk about the properties or characteristics of networks. We

will learn about the kinds of structures that can be used to distinguish and identify
different classes of networks.
So another question we would love to answer is, for example, you had a cold last week
and you want to know what you should blame for this. How do we infer how a virus
spreads on a network? How can we infer how neighbors in the network influence each
other? These are also questions that we will discuss in this module. Also, you might
not remember who exactly you have met with last week, maybe we can recover the
underlying network describing who interacted with whom just from data on how the
virus spread. Similarly, we might be able to recover the links between websites just
from data on how often each website is visited. Learning the underlying network from
the data generated by a process on the network is another theme in this module.
6
Finally, we want to be able to make predictions either regarding the future or for nodes
that we did not observe using the underlying network structure and data on a
particular process.
⇒Was a certain person infected by the virus or how likely is she to catch it in the
future?
How likely is a new link between two websites going to be used?
We do not know the answer because the link does not yet exist, but a prediction will
help us to decide if we should add it.
Whether we're interested in the spread of the virus, an idea, or a tweet. The same
basic ideas will allow us to understand what's going on to make predictions about
unobserved quantities.

1.2 Definitions and Examples of Networks

Uploaded by

Copyright:

Available Formats

1.2 Definitions and Examples of Networks

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

1.2 Definitions and Examples of Networks

Uploaded by

Copyright:

Available Formats

Definitions and Examples of Networks

In order to study networks, it will be useful that we are

Mathematically, a network is an object called a graph. It

Another example of a physical network is the road network.

Another example is a gene regulatory network. Here the vertices

In the recommendation system, relationships between items or

In Facebook friendship is an undirected relationship. So you'd use an undirected graph.

Also, citation networks are usually DAGs, since a paper

The information in a recommendation system can be

More generally, we will talk about the properties or characteristics of networks. We

You might also like