Using InfraNodus, you can get an overview of any YouTube video content by creating a knowledge graph from its subtitles. You can then use this graph to understand what the video is about, jump to the relevant parts of the content, and generate interesting research questions that help you develop the discourse further.
To try it out, find a YouTube video with subtitles, copy its link, and paste it in the "Add New Text" field in InfraNodus (or go directly to infranodus.com/import/youtube).
Make sure you Select "Analyze YouTube Subtitles" option, then choose whether you want to analyze only the content or also extract the entities in the text, and click "Visualize". After a few seconds, you'll generate a knowledge graph that shows all the main concepts and their relations, which you can use to get to the most relevant parts of the video, analyze its content, and generate new ideas on top of existing discourse.
Here's a brief tutorial that shows how it works:
What you can do using InfraNodus:
1. Get an overview of the main topics for a video
Once the video subtitles are visualized, you will see a graph with the main concepts that are found inside the video. The words used in the video are the nodes, and their co-occurrences are the relations. The more "influential" terms are bigger on the graph, while the ones that are used frequently in the same context will have the same color. Those concepts will be grouped into topical clusters and shown in the "Main topics" panel of Analytics:
For instance, for this interview between Tucker Carlson and Vladimir Putin, we can see that the main topics were the Ukraine Russia relations, Russian Emprire (and history), World War, and Military Power.
2) Entity extraction — build a knowledge graph from YouTube content
By default, InfraNodus builds a graph from all the concepts found in a video. However, you can set it to extract entities and to build a graph only from the relations between these entities:
As a result, you will get a more sparse graph, which may be easier to analyze for long-form interviews:
For instance, instead of two words for "united" and "states" or "soviet" and "union", we get "united states" and "soviet union".
This sort of analysis is interesting when you want to have fewer entities shown on the graph and extract structured data from YouTube content.
3) Jump to the most relevant parts of the video
Suppose, you saw the "nuclear weapon" node in the graph above and want to jump to the part of the video where Putin talks about this topic. Just click on the node, you will have the statements filtered on your left, which will contain the link to the timecodes where Putin talks about this subject. Click on a link and you'll open the video in a new window at exactly this moment:
4) Generate a summary of the content
You can summarize the whole video or only some parts. To do that, just click "AI Summary" or "Summarize Visible" button at the top menu and you'll have GPT-4 generate a summary of the visible statements for you.
This is useful not only for the whole video, but also for selecting a certain part of the video (containing "nuclear weapons") and generating a summary of all the statements that contain this topic:
5) Identify gaps in the video's content and generate new ideas
You can also use the topical structure of the knowledge graph to identify the gaps (blind spots) between different ideas presented in the video. You can then use this gap to generate new research questions and ideas using AI that help you bridge those topics in a new and interesting way.
To do that, go to Analytics > Blind Spots
Read more about it on Identifying Structural Gaps in a Discourse
Comments
0 comments
Please sign in to leave a comment.