Text Analysis Workflow: Get an Topical Overview and Find Content Gaps

Below you will find a basic text analysis and diagnostics workflow based on the InfraNodus research framework. Its goal is to provide an overview and encourage cognitive variability to generate multiple perspectives. The interactive knowledge graph enables you to alternate between different scales of perception (zooming in/zooming out) and intent (focus/exploration). As a result, you can understand both the surface-level and deeper layers of any discourse, uncovering the main topics as well as the structural gaps within.

This workflow is available on every InfraNodus visualization through the Workflow menu at the top right. As you go along the workflow items, you can mark them off as completed or click "Show" to see which menu elements correspond to each of them.

Text Diagnostics Workflow

Here we follow the steps in the text diagnostics workflow to analyze a discourse:

1. Add or import a text
copy and paste your notes, upload files, or import from external sources

2. Get a visual summary of the main topics
reveal the main topics inside the text, see how they connect, generate topic-based summaries

3. Reveal the content gaps
find what's missing inside the text by focusing on the gaps between the topics

4. Explore the text using the most relevant concepts and topics from the graph
use the interactive graph to see where the main concepts and topics appear in original text and explore them in more detail

5. Generate and save your insights
use the AI panel to generate insights based on the graph selections and write down your ideas into Project Notes

Here is the explanation of each step in more detail.

1. Add or Import a Text

You can copy and paste your text, research notes or upload a research paper or a collection of documents, or import a text from many of the external sources available in InfraNodus (e.g. a YouTube video transcript or your Evernote / Obsidian notes):

2. Get a Visual Summary of the Main Topics (High-Level Ideas)

Objective: after this step you should get a good understanding of the high-level topics contained in this document and get a concise visual summary of the main concepts.

InfraNodus will build a knowledge graph from your text. The concepts (or named entities, if you choose this processing option) are the nodes, their co-occurrences in the same paragraph are the relations.

⇢ a. Click Analytics > AI insights to see the main topics and AI-generated summary
⇢ b. Check the graph to see the main concepts and high-level ideas (use the toggle to turn them on). See a list of topics on the Analytics > Main Ideas panel
⇢ c. Use the AI Insights panel to generate summaries for the topics identified in the text

You can use the graph to get a visual overview of the content. The most influential concepts (based on betweenness centrality, not frequency) will be bigger in the graph. The concepts that frequently occur together will be closer to each other and have the same color (based on the Force-Atlas layout and community detection algorithms):

You can inspect the graph to see how well-connected the main high-level ideas are. If a topical cluster is too much at the periphery of the graph, you might want to bring it back in. If its too 

InfraNodus will apply network analysis metrics (community detection) to reveal the main topical clusters in the text and auto-generate names for them.  It will also use its underlying GraphRAG AI to generate a general summary and a research question to help you get a better understanding of the content. You can see these insights in the Analytics panel.

The Main Ideas provide more detailed information about the topical clusters and you can generate an AI summary for each of those topics as well, to get a more structured understanding of the content:

If you want to see the main topics on the graph, you might want to select AI: Show Categories to reveal high-level ideas that the main topics belong to.

Moreover, you can generate a summary for every topic, so that you know what each topic is about:

3. Reveal the Content Gaps 

Objective: after this step you will know what's missing in this text and know what questions to ask to develop this discourse further.

Some of the topical clusters or concepts inside the text are not well connected and produce structural gaps in the network. These gaps can be used to generate insight because they connect the ideas in new ways that are still contextually relevant. Select Analytics > Gap Insight panel to reveal the structural gaps in this discourse. 

⇢ a. Click Analytics > AI insights to see the content gaps in the discourse
⇢ b. Click AI: Question or AI: Idea to generate an idea to bridge this gap

Think of a research question or an idea that can bridge the structural gap. Use the built-in AI for inspiration, if needed (we recommend using it to generate questions but to answer them yourself). Click "Show Another Gap" to load more structural gaps. Bridge them with your ideas or AI.

4. Explore original text using selected concepts and topics

Objective: after this step, you will have a deeper understanding of the concepts and topics that are relevant to your specific interests.

After a high-level overview, you can zoom in and explore the original content in more detail. InfraNodus graph allows you to read the content in a non-linear way using the graph.

The most influential concepts are shown bigger on the graph. Relations will show which tend to co-occur in the same context. Click these concepts in the graph to see the original statements where they appear.

⇢ a. Click the nodes in the graph to see the statements where they appear
⇢ b. Click the topics in the analytics panel to filter the statements that belong to this topic

Use the AI Summarize button to summarize the content of those visible statements if you don't want to read through everything manually.

Select some topics in the Analytics panel to see the groups of concepts that belong to them and to understand each one of them better. The statements panel will automatically show the statements that belong to this topic only, so you can see the parts of the text that touch upon this particular topical cluster.

Use the AI: Summarize Selected Topic button to generate an AI summary for that topic:

5. Generate and Save Your Insights

Additional Steps for Advanced Text Analysis:

Here is what you can do to analyze deeper, latent topics inside the text. The core of this approach is "slicing off" the layers of the knowledge graph that you already understand in order to get to the underlying ideas

6. Reveal Underlying Ideas to Explore Nuance & Peripheral Ideas

Objective: after this step, you will have an understanding of underlying ideas in this text that are not easily visible at the first glance

 

Remove the most influential concepts from the graph to reveal the underlying ideas around them. This will help you extract the nuance and explore the peripheral topics that make a discourse special and can potentially serve as low-barrier entrance points. You can use the Analytics > Reveal Underlying Ideas button or select and hide the biggest nodes directly in the graph.

 

⇢ a. Click the nodes and then click the "eye" button to hide them from the graph
⇢ b. Use the Analytics > Content Gaps > Conceptual gateways to reveal the latent nodes

 

7. Reiterate: Overview → Zoom In → Gaps → Ideas

8. Save the Graph State and Reset the View

7. Import More Data

You can add more files or import more data into the same graph to explore more ideas. Alternatively, you can create more graphs and then compare them to see how they are similar or different.

8. Finalize Your Project Notes Insights

Once you're done, create a summary or an outline of your project notes using the built-in GPT AI.  You can use it to write a new article or generate a prompt to enhance your AI workflows.

9. Share Your Results

Download the PNG / SVG (high-resolution) imagine of the graph, export your notes and / or AI insights with the final summary and write / generate a new article or create a new graph.

...

Introduction to the InfraNodus Interface