Thanks to Chan Yee Hui for this write up of his recent 1-day workshop at RMIT University on intermediate network analysis using NodeXL and Netlytic. The content was tailored for teaching PR and communication students.
- The focus this time was on Facebook and YouTube and less on Twitter. In Vietnam, Facebook and YouTube are still largely the dominant social networks.
- For Facebook we covered the following scenarios:
- Analysing a fan page as a page admin. As page admin we have better access to user’s data such as username of each post’s authors, each user’s first, last and username.
- Examples visualization:
Posts responses of Dataviz Fan Page
- Node Size = Total Likes. We can quickly see which posts garnered the most likes
- Post that are liked/reacted by more users has more edges
- G1 = posts that do not get any reaction/likes
- G2 and G3 generate more interests and likes from users.
- From a glance:
- Posts that promoting tech or news generally get less reactions and interests
- Posts about events organized by Dataviz generally get more attention
User interaction from posts in Dataviz Fan Page
- Node Size = Betweenness centrality.
- Edge size = Edge’s weight (how many count of relationship between the 2 actors. The more interaction (like, comment), the thicker the edge)
- User that shared more common reaction with others will increase in size
- Each segment centered around a key user.
- We can find out the most important players in this network
- Analyse the relationship between pages liked by a particular Fan Page, covering distance up to 2 degrees. This allows us to analyze what are the pages that are “liked” by a page, and the pages that are “liked” by all these 1st-degree page, and so on up to 2 degrees.
- The higher the degree, the size of the graph will increase exponentially. We discussed with the class on how to start small, experiment, and gradually increase the distance to reduce processing time.
- Sample visualization includes:
A 1.5 degress page-likes network for official Mercedes-Benz fan page
- The last scenario we covered for Facebook are the relationship between posts and their authors within a Fan Page, as a non-admin user. This allow us to analyse any Fan Page with limited end-user data.
- Example visualization:
- This is visualization of comments made within a Fan Page, and the posts’ relationship with each other.
- Each node represents a post, and if 2 posts shared the same commenter/reactor, a line will be drawn between them.
- We can see different segment of comments and their top keywords
- This allows us to analyse comments that interact closely together (clustered segment means those comments/posts are closely connected)
- We can analyse comment that is more popular or garnered a lot of attention
- We also generated a few time series to show the trend and frequency of comments and replies in the network
Finally for our YouTube analysis, we covered 2 main scenarios:
- We did a keyword search to get all related video, and analyse the comments made from these videos..
- We used the 2019 Gillete #Meetoo campaign as our use case
- We can see a few segments from the comments. Based on the top words on each segment, we can conclude that the campaign did not generate a positive response
- The size of each node is determined by the video’s view. The more views the larger the image
- The size of the edge is determined by the edge’s weight (how many common commenter between these 2 videos)
- We can quickly identified the most viewed video, the video that has the most common users (thick lines)
The second scenario is to use Netlytic to analyse a single video and its comments. We used the Tesla latest model launch video as an example:
With Netlytic we demonstrate how easy it is to generate high level insights and basic keyword analysis for a YouTube video.
The participants went on to create some interesting experiments with their own Facebook pages and videos which yielded insightful results. We look forward to our next workshop with RMIT!