skip to Main Content

NodeXL Pro Tutorial: Introduction to “TikTok Post Networks”

Discover how to analyze and visualize “TikTok Post Networks” using NodeXL Pro in our latest tutorial that will guide you through the process of collecting TikTok data with the versatile Zeeschuimer tool, importing the data into NodeXL and analyzing the network structures to uncover insights into user interactions and content trends.

EURO 2024 network graph

Situating TikTok Networks

TikTok is a prominent social media platform characterized by its focus on short-form video content. Since its start in 2016, TikTok has experienced exponential growth, amassing over a billion active users globally facilitating a unique ecosystem for the dissemination and viral propagation of multimedia content, encompassing various genres such as entertainment, education, and social commentary.  Its diverse user base and high engagement levels produce huge amounts of social network data, thus providing the grounds for studying digital communication patterns and user behavior in contemporary social media landscapes.

Just like any social media platform, TikTok provides its users with a classic set of interactive options that form networks – like mentioning another user in a post, liking a post, commenting on a post or replying to a comment. In an ideal world we would be able to gather all of this data into one network analysis. In real life it is not that simple, but we are working on it!

In the following we will look at “Post Networks” which concentrate on the contents of original posts around any search term or user – these will not contain the comments and replies. The comment and reply level will be addressed in a different tutorial soon.

Post Networks” offer at least two different levels of analysis worth exploring in detail:

  1. User-mentions-User Network: When a post author mentions another user in a post, one network edge is created. This results in a classic uni-modal user network in social media. This approach allows the detection of communities (clusters) and influential users as well as the analysis of shared contents.
  2. User-mentions-Hashtag Network: When a post author makes use of a hashtag in a post, edges are created from the post author to the hashtag. These edges then result in a two-mode network with users as Vertex 1 and hashtags as Vertex 2. The main benefit is the emergence of content clusters in which users form around commonly shared hashtags.

Tutorial requirements

Getting data

Zeeschuimer is a versatile tool that allows (manual) data collection from various social platforms, including LinkedIn, Instagram, X (formerly Twitter), and more. It also offers two types of TikTok data sets – Posts and Comments. In this tutorial, we will concentrate on collecting and analyzing TikTok Posts data with NodeXL.

  1. Install the Zeeschuimer tool.
  2. Open a TikTok page with a search term of your choice and collect data by simply scrolling down the posts. You will receive a maximum of about 300 posts per scroll. These posts are a mixed selection of currently trending and popular videos. You can do multiple scrolls. A mouse function that automatically scrolls to the bottom is very helpful here!
  3. Download the .ndjson file from the Zeeschuimer tab.
  4. Open this page to drag and drop your .ndjson file into the central window.
  5. Save the file as a .csv file to your machine.
  6. Import the .csv file into Excel via Data > Get Data > From File > From Text/CSV. When formatting in UTF-8 the emojis in the text columns will remain intact.

NodeXL Pro data import

To open the NodeXL Pro Import From File data importer select NodeXL Pro > Data > Import > … From File:

  1. Select your Excel file via the Browse… button.
  2. Click Read File(s). You can now manually configure the settings in steps 3 to 6. Or you can continue at step 7.
  3. Customize the data table.
  4. Check-mark columns from your file to either be imported as Edge columns, Vertex columns or both.
  5. Construct network edges within the “Build an edge between” section. E.g. you can create an edge with the relationship “Mentions User in Post” from the Name/Author (Vertex 1) column to each mentioned user (Vertex 2) in the text/message column “body” with “@” used as identifier. You find more details about potential edge relationships below.
  6. Define the Edge Relationship types and “Add” them to the table on the right. Repeat this step for each edge relationship you have defined.
  7. Optionally save your configurations via “Save Config” from steps 3 to 6. Or load the configuration file provided in the zip-file above “_config zeeschuimer_tiktok_posts.NodeXLImporter”. The edge creation process is explained in the table below.
  8. Click OK to start the data import. This process may take a few minutes depending on the table size.

Edge creation

The above Config file is just a proposal for the creation of three edge types. You can modify or add further edges as you like:

  1. User mentions User in Post: When a post author mentions another user in a post, one edge is created. Multiple edges for one post are possible.
  2. User mentions Hashtag in Post: When a post author makes use of a hashtag in a post, edges are created from the post author to the hashtag. These edges then result in a two-mode network with users on the one hand and hashtags on the other. You can easily remove these edges from the analysis by setting the edge column “Visibility” to “Skip”.
  3. User posts: A self-loop from the author to the author itself is created to ensure that all posts are included in the analysis even if no edges are formed. That means Vertex 1 and Vertex 2 will be the same here.

The following table shows the setup of the above Config file. The Source Vertex is the column that contains the post author. The Target Vertex is the column which is used to find mentioned users and hashtags:

NetworkRelationshipSource VertexTarget VertexConfiguration
User-UserMentions in PostauthorbodyEach word that begins with @
User-HashtagMentions Hashtag in PostauthorbodyEach word that begins with #
User-UserPostauthorauthorOtherwise

Network and Content Analysis with NodeXL Pro

After data import we need to prepare the data for analysis:

  1. Navigate to the Vertices spreadsheet. Find column “author_avatarThumb” and copy and paste the contents to column F “Image File”. This way we can display the images of the post authors in the network map. Note that the links to the user profile images will expire after a few days and updating the links is currently not possible.
  2. Navigate to the Edges worksheet and find the Relationship column. Remember that we have created a mix of relationships. In column G “Visibility” you can set the value to “Skip”, to disregard edges in the upcoming analysis and focus on a certain edge type. First we will look at the User-mentions-User network, then at the User-mentions-Hashtag network. So to begin we recommend you set the edge type of “Mentions Hashtag in Post” to “Skip”.

Next step is Task Automation – the key feature of NodeXL Pro. This feature allows the automation of all tasks of the network analysis and visualization. Learn more about data recipes and task automation here. In the next step we will import customized NodeXL data recipes that were customized for this tutorial.

  1. Download this .zip file, unzip it and save the folder to your machine.
  2. Select NodeXL Pro > Options > Import…
  3. Navigate to the .NodeXLOptions file (a.k.a. data recipe) from the zip file mentioned above: zeeschuimer_tiktok_posts.NodeXLOptions
  4. Open the “Automate” dialog box via NodeXL Pro > Graph > Automate and select “Run”.
  5. Wait for a short while until Task Automation is finished.

The resulting network map may look something like this:

Initial tests have shown that most Tiktok Post Networks have a similar shape: The network is very sparse resulting in one large “Brand” cluster of isolates which are users who do not mention any other accounts in their posts. And also there are a few small connected components. This network will look much different when comments and replies are added to the data set.

The User-Hashtag Network

By analyzing how users connect to hashtags, it is possible to identify clusters of users and hashtags, indicating communities or groups with shared interests or themes.

  1. To create this network, first save the workbook, then save it to a new file.
  2. On the Edges worksheet remove the “Skip” from all edges.
  3. Optionally set a “Skip” in the Visibility column for the edge relationship “User mentions User in a Post”.
  4. Then run Task Automation again. This time you may want to use the data recipe “zeeschuimer_tiktok_posts – user-hashtag network.NodeXLOptions” which runs the same steps of the network analysis but contains a few changes in the layout.

The resulting network map may then look something like this:

In comparison with the previous network graph, the user-hashtag network contains many clusters. Take a look at the most frequently shared words in the posts which act as a group label at the top of each cluster in the map. These are good pointers to identify topics of interest and discussion.

You find many more examples of such networks in the NodeXL Graph Gallery when you enter #nxltiktok into the search bar.

Questions?

Please send an email to info@smrfoundation.org.

Back To Top