BLOG 6 | Module 3

 Week 6 | Module 3: Social Media Analytics/Network Analysis

In this module we explored networks, network visualizations, their theory, tools and uses.
We also looked at the quantitative metrics, properties and layouts of networks and discussed how to use and interpret the analysis and the interactions in the networks.

Lecture 11 introduction to networks

Network, also called graphs are collections of entities and relationships among them. They consist of edges and vertices which represent relationships and entities respectively. Networks are very useful for various purposes some of which are information diffusion, disease propagation, and relationship formation. In a nutshell they help us study complex systems and their behavior.
Primarily there are two major types of networks
Single mode networks, where there are only one kind of nodes
Diving further into single mode networks:
               Single mode networks have only one type of vertex
               Single mode edges can be either directed/undirected and weighted/unweighted
Two mode networks, where you can more than two kinds of nodes
               under two mode networks vertices represent two different kinds of entities
               they can be either directed or undirected
Network visualizations
Network visualizations are a very important tool that can help us understand the communities and their interactions in a network. For directed networks can be represented in two ways using an edge list and by using adjacency matrix. Undirected networks can be represented using the same ways adjacency matrix represent symmetric relationships. Networks can also be divided into ego networks and whole networks.

Lecture 12 Introduction to network visualization

This lecture was about understanding the purpose of visualizations, network visualizations with different kinds of layouts.
I'm a visual learner so this segment in particular was very interesting to me. While information visualization has many purposes, they can be divided into three broad categories: Exploration, Communication and Understanding.
There are various types of network layouts which we can use for our network based on our needs. The few layouts we looked at are:
Force directed
It minimizes the chance of collision between nodes. The nodes and their connections are laid out in a way that they are distinct and do not crisscross each other (very less if any)
Geographical: Maps nodes to physical locations
Circular: these can be in two ways either the nodes and their relationships are laid out in single circles or they are laid in concentric circles with relationships between layers
Clustering: these help identify connected components in large networks
Hierarchical: these help identifying relationships among nodes in a hierarchical manner

Lecture 13: Network properties

In this lecture we looked deeper into the structural properties of networks. These properties are usually mathematical, and they help us understand our network and its components better. We can use these property values to make smart decisions for our goal or to study the interactions and propagation of the network
The various properties that we discussed are:
Degree centrality: degree of a node often called the degree centrality of the node is the number of links that are incident on the node. This is used as a measure of the node’s influence/ popularity in a network.
Paths and shortest paths: a path between two nodes is a sequence of non-repeating nodes that connects the two nodes. The shortest path is the path that connects the two nodes with the shortest number of edges. This is useful in cases where longer paths are desirable or not desirable
Betweenness centrality is the number of shortest paths that pass through a node divided by all the shortest paths in the network. It is used to determine which nodes are more likely to be in the communication path between other nodes, and where the network would break if a particular node was disconnected
Closeness centrality is the average length of all the shortest paths from a node to every other node in the network. This is a measure of reach and is useful in cases where we try to understand information dissemination.
Eigenvector centrality is proportional to the sum of the Eigenvector centralities of all nodes that are directly connected to it. It is the measure of the importance of a node, well connected nodes will have high Eigenvector centrality value
Reciprocity is the ratio of the number of relations with are reciprocated. These are only for directed network and indicate the degree of mutuality in a network
Density is the ratio of the number of edges over the total number of possible edges between all pairs of nodes. It is the measure of how well connected a particular network is
Clustering Coefficient: A nodes clustering coefficient is the density of its neighborhood. These help determining the sub-communities within a network.
We also looked at a few distance metrics: Longest path between two nodes in a network is called the networks diameter. Diameter is the indicative of the reach of the network.
Connected component: In undirected networks, a connected component is a sub-graph in which there is a path from any node to any other node in the sub-graph
Strongly connected component: If there is a directed path from every node to every other node in a connected component it is called a strongly connected component. It only applies to directed network.
Weakly connected component are those subgraphs where there is a path between two nodes only if we ignore the direction of the edges.
Bridge is an edge which when deleted increases the number of connected components.
Clique is a fully connected component in a graph

Readings

Thinking in network terms

This article by Professor Albert talks about networks, the various questions that come with network analysis and various use cases of major companies and departments that have used network analysis to their advantage.
It essentially talks about how we can use data to control systems, describe and quantify properties of a network mathematically and if it could also help us predict the behavior of the network system. we looked at how network analysis evolved over various stages: thinking of networks, calling it 'human dynamics', adding the science part to it (which is primarily what we focus on).
It highlights that we live in an economy of information and interconnectedness. How everything we use the phone, the internet, or any other devices we use, is essentially a reflection of our needs, communications needs, and it depicts our behaviors. The major scientific challenge to this pertains to the extraction of answers to our questions from this vast amount of data that is being recorded.
Data is a gold mine for science in todays world but factors like accessibility to technology and legal restrictions often skew the data and the information it is trying to represent. The article ends with something I agree with, we want to unlock potentially important information from various private resources that binds the scientist hand and to do so we could identify some social consensus by which data can be shared with users who could potentially benefit from it.

Explore your LinkedIn network visually with InMaps

In this article we explore how to work with interactive network visualization using InMaps. InMaps is a tool that helps visualizing your professional network which we can then analyze and use the information to our leverage.
This includes finding job opportunities, professional advices, gathering insights etc.

Social Network Analysis

In this article we look at social network analysis, what is its theory, applications and economic implications
Social network is a social structure made up of individuals that are connected via some sort of relationship. These structures can get very complex in their graphical representations, as they can have many kinds of ties and levels in them. Social network analysis has become a key technique in modern sociology.  This article defines various properties of a network that we looked at in our lectures and it also defines the mathematics of a network graph which also lines up very well with our lecture. Something new and interesting for here were exploring social network analysis tools and their features. A few of them are AllegroGraph, AutoMap, CFinder, Commentrix, and CoSBiLabGraph.
We also had Networks, Crowds, and Markets: Reasoning About a Highly Connected World book as a reference for this week. This book takes a scientific perspective to understand networks and its behaviors. Throughout its parts and chapters, it takes examples from economics, information science, mathematics, and sociology; and seeks to address questions about how these worlds are connected to each other.

Articles

https://www.latentview.com/blog/a-guide-to-social-network-analysis-and-its-use-cases/

I chose this article for this module as it lines so well with our readings and lectures, and it also offers detailed use case examples of how we can use social network analysis to our advantage. The use cases range from HR optimization to finance and fraud detection. My personal favorite was the fraud detection use case. I love how they broke down the steps that help us identify a fraud ring.

Source: https://www.zaloni.com/resources/blogs/detecting-financial-fraud-social-network-analysis-graph-database/

Let me know which one you found the most interesting in the comments below!

Since we also looked at data visualization tools, I thought I should include this in this week’s articles. This article, lists the top 7 network analysis tools in the market today.
https://analyticsindiamag.com/top-7-network-analysis-tools-for-data-visualisation/