Presented at ILTA EdTech 2017, Sligo, Ireland
Supporting posthttps://mashe.hawksey.info/?p=17538
Patterns are left behind. Whether it be replies to a discussion forums, interactions on social media or ingredients in cocktails links can be made and the data used for actionable insight. Network science is one approach that takes these seemingly complex connections and through the use of mathematical methods make it easier to understand. Network science is a well established discipline and it’s origins can be traced to 1736 and the work of Leonhard Euler. The area of social network analysis is a more recent development established in work by Moreno and Jennings in the 1930s. Accessibility to affordable computing in the 1990s combined with data from early social networks like IRC has led to an explosion of interest in social network analysis. This has continued with the emergence of social networking sites like Facebook and Twitter combined with accessibility to the underlying data. The use of network science and social network analysis within educational contexts has seen similar growth. The emergence of ‘Learning Analytics’ as a field of study has highlighted how data can be used to enhance learning and teaching. With social network analysis we can take seemingly complex relationships and making them less complicated. Common applications of network analysis in this area include: identification of isolated students within group activities; identification of people or concepts which are ‘network bridges’; clustering of categorisation of topics; plus numerous other applications.
This presentation is designed to be an introduction into network analysis allowing delegates the opportunity to understand the underlying structure of the graph as well as some of the tools that can be used to construct them. The session will begin with an introduction to key network analysis terms and go on to introduce some of the tools and techniques for social network analysis, specifically looking at how data can be collected and analysed from Twitter using tools like TAGS and NodeXL.
3. Making the complex less complicated:
An introduction to network analysis
Martin Hawksey
@mhawksey
#iltaedtech17
http://go.alt.ac.uk/iltaedtech17-networks
This work is licensed under a
Creative Commons
Attribution 4.0. CC-BY
mhawksey
13. alt.ac.uk
Examples
Bakharia and Dawson (2011) SNAPP: A Bird’s-eye View of Temporal Participant Interaction
https://www.slideshare.net/aneeshabakharia/snapp-learning-analytics-and-knowledge-conference-2011
Learner Isolation Facilitator Centric
@mhawksey
14. alt.ac.uk
Examples
Bakharia and Dawson (2011) SNAPP: A Bird’s-eye View of Temporal Participant Interaction
https://www.slideshare.net/aneeshabakharia/snapp-learning-analytics-and-knowledge-conference-2011
Non Interacting Groups Facilitator Bias
@mhawksey
17. “
alt.ac.uk
Graphs can be a powerful way to represent relationships
between data, but they are also a very abstract concept, which
means that they run the danger of meaning something only to
the creator of the graph. Often, simply showing the structure
of the data says very little about what it actually means, even
though it’s a perfectly accurate means of representing the
data. Everything looks like a graph, but almost nothing should
ever be drawn as one. Ben Fry in ‘Visualizing Data’
@mhawksey
24. alt.ac.uk
Key points
◊ Getting to this and you are over
80% or the way
◊ There are a lot of very
knowledgeable people in the
community willing to help
◊ Go explore … and have fun
@mhawksey
25. alt.ac.uk
Getting Social Network Data
◊ Using Twitter as a data source: an overview of
social media research tools (updated for 2017)
◊ Twitter: How to archive event hashtags and create
an interactive visualization of the conversation
@mhawksey
Abstract
Patterns are left behind. Whether it be replies to a discussion forums, interactions on social media or ingredients in cocktails links can be made and the data used for actionable insight. Network science is one approach that takes these seemingly complex connections and through the use of mathematical methods make it easier to understand. Network science is a well established discipline and it’s origins can be traced to 1736 and the work of Leonhard Euler. The area of social network analysis is a more recent development established in work by Moreni and Jennings in the 1930s. Accessibility to affordable computing in the 1990s combined with data from early social networks like IRC has led to an explosion of interest in social network analysis. This has continued with the emergence of social networking sites like Facebook and Twitter combined with accessibility to the underlying data. The use of network science and social network analysis within educational contexts has seen similar growth. The emergence of ‘Learning Analytics’ as a field of study has highlighted how data can be used to enhance learning and teaching. With social network analysis we can take seemingly complex relationships and making them less complicated. Common applications of network analysis in this area include: identification of isolated students within group activities; identification of people or concepts which are ‘network bridges’; clustering of categorisation of topics; plus numerous other applications.
This presentation is designed to be an introduction into network analysis allowing delegates the opportunity to understand the underlying structure of the graph as well as some of the tools that can be used to construct them. The session will begin with an introduction to key network analysis terms and go on to introduce some of the tools and techniques for social network analysis, specifically looking at how data can be collected and analysed from Twitter using tools like TAGS and NodeXL.
According to some in the last two years we have doubled the amount of data stored. We are in unprecedented times. Previously the census may have recorded my name, where I live and occupation, but now so much of our activity is recorded, digital footprints are everywhere. This is a huge ethical concern but I believe there are situations where we can use this data is a positive way to improve learning, life and society. But how can we make sense of these complex interactions…
Digital paw prints left behind
One solution is network science. This isn’t a new field of study although the emergence of affordable computing and access to data has accelerated it’s growth. From a social perspective Jacob Moreno is often cited as the inventor of sociagrams. These graphs show the relationships between groups of school pupils. Before going further with what’s possible using network analysis some terminology/concepts …
Some terminology … a point is a node (or vertex) … connections between nodes are called edges or links. In this example the edges are directed. For example, on twitter I can follow you but you might not follow me back, but on Facebook two people are friends so it’s undirected.
Nodes can be anything not just a person. They could be an email, discussion post…
Now for the science bit. Having created networks we can makes and use different measures. For example, we can count the number of connections a node has. This is called ‘degree’.
The degree of a node calculated by the number of edges that are adjacent to it. So by ranking each node within a social network by degree, we can distinguish which individuals have the most connections (Figure 6). Source http://www.richardingram.co.uk/2012/12/visualising-data-seeing-is-believing/
There are a number of measures we can use and other the years various algorithms have been developed to analysis networks providing methods looking density within graphs, clustering, layout and more. For example, often you can calculate ‘betweenness centrality’.
Betweenness Centrality measures how often a node appears on the shortest paths between nodes in a network. So by ranking each node within a social network by betweenness centrality, we can distinguish which influential individuals have the most connections across distinct community clusters (Figure 7). Source http://www.richardingram.co.uk/2012/12/visualising-data-seeing-is-believing/
You might have head of some of the other network analysis tools. Have you heard of PageRank? PageRank was the algorithm developed by Larry Page and Sergey Brinn which they later used to found Google. Whilst the algorithm Google uses for listing search results has changed the original research is still used in network analysis
The PageRank graph is generated by having all of the World Wide Web pages as nodes and any hyperlinks on the pages as edges. The edges are further characterized as weak or strong edges by weighting the edges. Pages that are linked by more credible sources such as CNN or USA.gov sites have higher weightings for the respective edges. Thus, if we compare two sites with the same number of edges. PageRank will give the site with more links to credible sources a better rank. Source: http://blogs.cornell.edu/info2040/2011/09/20/pagerank-backbone-of-google/
Threshold concept for me was that a lot of software builds networks from a paired list. So if you can find ties between thing A and thing B you can start creating networks.
So how can we use network analysis in learning. Here are some examples
This is one of the first graphs that got me interested in network analysis. It was produced by Tony Hirst at the OU who had just started exploring network analysis techniques himself. The graph shows the connections between Twitter screen names for a community hashtag. Looking at this graph one of the personally powerful and motivational revelations was to see I was part of a community. If you consider a medium like twitter it can be hard for you to get a sense of where you are in a community and who else is part of it. Seeing myself exist in the graph gave me a sense of place but it also let me see who else in the community I was close to but not connected with, or even discover people on the other side.
This type of network graph is often called a hairball … or as I call it the big ball of timey, whimy, whibbly wobbly stuff.
One criticism of graphs, all graphs, not just network graphs, is often they are only truly meaningful for the creator of the graph. So we’ve taken something complex and made it more complicated. One thing to remember within network analysis is whilst the research paper or blog post contains a static image it is through the active exploration that the real answers are revealed..
Fortunately there is a long list of tools, many open source, designed for the exploratory analysis of networks.