#gephi analysis

[aianalysis-blog]

Gephi Analysis of #VirtualReality

Network Analysis

Figure 5. Who Mentions Whom. 

Figure 6. Who Replies to Whom.

These two images were also from Netlytics, based on data I analyzed. It shows who mentions whom and who replies to whom on Twitter. In Figure 5, who mentions whom, you can see a lot of big nodes in the centre of the image and a lot of little nodes surrounding the big nodes. In Figure 6, who replies to whom, it clearly has a better structure of the connections between the people who reply to each other. In Figure 5 it’s all blended into each other, while in figure 6, all the nodes are far apart from each other.

Gephi

I used the network analyzing tool ‘Gephi’, to see for myself what the connections between the hashtags I used are. The hashtags I used were as mentioned before #education and #virtualreality. I analyzed data on Twitter, to get a better picture of how the Twitter accounts and their tweets are connected with each other. I analyzed this data through Gephi with three types of network graphics, these are based on who mentions whom. These graphics are In-Degree, Out-Degree, and Modularity Class.

Figure 7. Gephi, Out-degree 

Figure 8. Gephi, In-degree 

Figure 9. Gephi, Modularity

[aianalysis-blog]

Gephi Analysis of #IOT

Network analysis

Figure 1.7 Gephi, modularity class 

Figure 1.8 Gephi, Out-degree class 

Figure 1.9 In-degree class

When I put the layout within the graph together with strong gravity, the two biggest nodes within and out degree became way more clear. Userexperienceu and rww are the two biggest nodes. Userexperienceu is Karl Smith that has loads of knowledge about currencies and AI and want to show this knowledge on his twitter with a lot of explanations how things should be done, he set up four currencies and he is the inventor of open network eco-system protocol which delivers smart-living ecosystems for houses. Rww is an information website about the internet of things and the latest news on it and interesting topics such as fintech and blockchain. As you can see in the in-degree class (which means measuring who gets listened to the most) They are the biggest blue dots and the network is very dense and connected with each other. In the out-degree graph (where you can measure who listens the most), you can see the yellow small dots but again the two big dots so it means that they also actually listen to each other a lot.

[aianalysis-blog]

Gephi Analysis of #DeepLearning

NETWORK ANALYSIS

Figure 1.5 shows the visualized connections about “who mentions whom”, where as figure 1.5.1 is a close up of the middle nods. To make this a bit clearer, I used Gephi.

 Figure 1.5 Who Mentions Whom

Figure 1.5.1 Who Mentions Whom Close Up 

Figure 1.6.1 #deeplearning [who mentions who indegree via Netlytic]

Figure 1.6.2 #deeplearning [who mentions who modularity via Netlytic] 

Figure 1.6.3 #deeplearning [who mentions who outdegree via Netlytic] In the above figures [1.6.1, 1.6.2, 1.6.3], we see a lot of users who mentions other users, for example by retweeting, whilst using the hashtag #deeplearning. This Gehpi graph has 554 nodes and 6340 egdes, which makes it a large data set. Some users with the largest connections are; @tsundu_mak, @arkangelscrap, @mikequindazzi, @evankirstel and @ronald_vanloon.

Figure 1.7 Who Replies To Whom

Figure 1.7.2 #deeplearning [who replies to whom indegree via Netlytic] 

Figure 1.7.3 #deeplearning [who replies to whom outdegree via Netlytic]

Figure 1.7.4 #deeplearning [who replies o whom modulartity via Netlytic]

In the above figures, we see the connections about who replies tot whom whilst using the hashtag #deeplearning. In this Gephi graph there are 20 nods and 15 edges, so it is not as big as the who mentions whom graph. But again, there are a few users which stand out more. These are; @tsundu_mak, @m_learningnews, @machinelearnonl, @dlandmlapp, @machinelearninz and @analyticbrigde.

[aianalysis-blog]

Gephi Analysis of #FinTech

By using Gephi as a tool of network analysis, I could certify the important people or organizations in this network and clearer connections between each twitter account. I looked for three types of network graphs;in-degree, out-degree and modularity.

Figure8. In-degree. Who mentions whom (Gephi)

In-degree measures the influence of accounts and it shows which users are the most mentioned accounts. This graph has several large nodes with many smaller surrounding nodes. In this network, three nodes with red colour;Fisher85m, evankirstel, mikequindazzi are the most influencial users with the #fintech in twitter. First, Fisher85m is the account that belongs to a person named Michael Fisher, an Analyst and Technology Evangelist - a person who inspires and convert businesses and individuals through interaction, content creation and communication about a technology. Evankirstel also belongs to a person, Evan Kirstel, a social media influencer in IoT, Data Security. Lastly, mikequindazzi is an account of Mike Quindazzi, a Managing Consultant at PwC.

Figure9. Out-degree. Who mentions whom (Gephi)

Out-degree centrality shows measurement of sociality. Therefore, by analyzing this graph, I could check who is the most active(communicative) actor in the discussions about #fintech on Twitter. Azurflorian, a company called “IPFC online” which focuses on Digital Marketing and E-Business, and bdsharmas, Bharat D. Sharma, an expert in Cybersecurity and fintech are the users that seem to highly reach out to others.

Figure10. Modularity. Who mentions whom (Gephi)

Modularity class describes how the one big network is divided into sub-networks. This graph shows what accounts are more densely connected to each other with the topic of Fintech on Twitter. This graph is a result of 86 communities (accounts). In this graph, there are four obvious groups of accounts that are connected to each other. First, Blue coloured group of nodes includes fisher85m, evankirstel, jblefevre60, antgrasso. By analyzing the main nodes of the blue group, I found out that the accounts in blue group have all belonged to an individual who are actively working at the finance and technology sectors. To be specific, antgrasso belongs to Antonio Grasso, a influencer and a speaker in the field of AI, blockchain and fintech. Second, nodes with red colour have mikequindazzi, azurflorian, enricomolinari, bdsharmas, 2vero as their main nodes. All these accounts were also belong to an individual except azurflorian. The group of nodes with pink colour was interesting because all of the nodes had a character of having a large number of followers and following. For example, ralexjimenez, which has 8810 followers and 665 following, is a twitter account of Fintech Strategist named Alex Jimenez and it is a verified account which indicates that the account of public interest is authentic. Syscoin, which is one of the most interesting nodes in pink group with 69818 followers and 5602 following, is a business on the cryptocurrency based on blockchain. By looking at those graphs through Gephi and Netlytics, I could find the answer of our sub-questions; What kinds of actors are involved in the debates on AI online? Based on the analysis that I did, most of the accounts with a power turn out to be the accounts that belong to individuals whose occupation is closely related to the finance or to the technology industry. Among them, there were lots of speakers who are actively giving presentations about the fintech sectors such as Michael Fisher, Evan Kirstel and Antonio Grasso. Since most active accounts are all experts in a technology field, who wants to tell the promising future of AI and Fintech, there were not enough messages about the risks of technology. Therefore, after I finished network analysis, the reason why there were only ‘Good’ comments in ‘Feeling’ category in text analysis (figure 5) became clear.

[aianalysis-blog]

Gephi Analysis of #MachineLearning

Graph 1: Modularity Class The first graph shows the modularity class graph. You can very clearly see that all the nodes are divided into groups, which means that these accounts mostly communicate in their circle of people. There’s not a clear node that’s much larger than the others, but the group of green nodes shows bigger than the others, which means that these accounts are more active than others.

Graph 2: In-degree The second graph shows the in-degree. You can tell that user mikequidazzi, and fisher85m are the most influential accounts in the tag. You can tell that this graph is quite dense, which means that there aren’t a lot of influencers in the debate. However, the jumping out nodes for users mikequidazzi and fisher85M make sense, as these are professionals in the AI field.

Graph 3: Out-degree This graph shows the most active communicators in the tag. Account machinelearn_d is the most active communicator, together with habitline. This makes sense as the machinelearn_d account is an actual account about all things machine learning. The graph is very dense with little nodes that jump out, so it is to conclude that machine learning does get talked about, but it’s not an active tag with a lot of daily communicators.