Keyword Clustering with AI Knowledge Graphs

Posted Wednesday, October 1, 2025.

Keyword clustering is a powerful approach to analyze the market discourse and identify the main topics and keyword combinations that are used by your target audience. Imagine being able to visualize the combinations of keywords that people use to search for your service so that you can better understand the main drivers behind the search intent.

The graph below shows an example of how it may work for the topic of "keyword clustering" itself. As you can see, when we remove the main search terms from the graph, we can see the context around them: people seem to be interested in the actual tools, so for us it's an indicator that making a free keyword clustering tool landing page would attract our potential customers:

 

The InfraNodus SEO app that we use above can identify those clusters not only from search intent, but also based on the co-occurrences of keywords in your competitors' websites and the search results. Unlike other tools that just show the tables, InfraNodus visualizes the results as a knowledge graph and applies advanced network analysis techniques to rank the keywords by their influence in the overall discourse, reveal the main topics and keyword combinations. This lets you quickly see the important patterns and use the graph to explore the keyword relations in more detail. You can also use the built-in AI to generate content ideas that bridges the gaps you discover.

This approach differs from the keyword clustering methods used in other tools, like Semrush or Ahrefs which group the keywords by similarity. InfraNodus uses additional data about the search intent and extracts the top search results to combine the data and identify the keyword clusters that are used by your target audience and competitors. You can then target the specific clusters or identify the gaps in the current informational supply and develop a unique content strategy that will cater to the topics that are in demand but not yet well-covered.

How Does Keyword Clustering Work?

Keyword clustering can be used to improve the topical authority of your website. When you cover the main topics and keyword combinations that are used by your target audience, you can build a coherent knowledge graph for Google as you will be linking all the important entities and topics together. This will help you rank higher for the main topics and keyword combinations that are used by your target audience.

 

For instance, the graph above shows the graph of the search results for a query. We remove the top two terms to see the context around them. We also apply the filter to only show the results from the top 6 positions. As we can see, "search intent" is an important topic, as well as "similar grouping" and "content strategy". Which indicates that when we create content around "keyword clustering" we should target these topics as well to make our content more complete and informative.

Once the content is created, it should be structurally linked in a cluster so that the search engines can establish a topical authority for the different combinations of keywords that we use. In our example, we would have a general page on the main search query and then additional pages link to from the main page that go into the details of the different topics. They would all link back to the main page to establish it as the main hub for the topic.

Search Intent vs Search Results

Another interesting approach is to compare a graph generated for search intent with the graph generated for the search results. We want to see keyword combinations people search for that they don't really find in the top search results. Finding those gaps will help identify content opportunities to bridge the existing informational supply with the search intent.

 

For instance, in the example for "seo tools", we can see that while people search for "keyword generator", this is not the keyword combination that comes up on the first pages of Google search results. If we create content related to "seo tools" that also targets "keyword generator", we'll signal to Google that we are covering an underserved topic and that will push our content higher.

Additional filters can be applied to show the keyword combinations with a high number of search queries per month. Built-in AI can be used to generate content ideas and article outline for these keyword combinations.

Content Gap Analysis of Keyword Clusters

Another interesting use case is to retrieve the content gaps in keyword clusters. If you are studying your competitors' websites, you would see the main topics they use. But there are some topics that they don't yet connect. If you target those gaps you will demonstrate to the search engines (and AI tools) that you're talking about the topics that are considered to be important for your target audience but in a way that your competitors do not. This will help establish your topical authority because you are adding something new to the existing discourse.

 

Knowledge Graphs vs Ahrefs and Semrush Keyword Clusters

Most of the SEO tools, like Ahrefs and Semrush, use their own proprietary algorithms to group the keywords by similarity. They do provide interesting insights, but their default results are a bit too generic and prectable. For instance, take a look at the screenshot below from Ahrefs: it shows us the main clusters for our query but those are quite obvious and much less specific than the data we can get from the knowledge graph.

 

InfraNodus uses a different approach by analyzing the search intent and the search results to group the keywords into clusters. The interactive graph lets you quickly remove the most obvious terms to see the layer of more specific keyword combinations hidden beneath. This approach is more transparent and allows you to manipulate the graph data to get to the more nuanced insights by removing the top nodes and unimportant terms.

In InfraNodus, we can remove the main search terms that take too much weight and get to the more nuanced topics. For example, if we export a list of keywords generated by Ahrefs in a CSV file and then visualize it with InfraNodus, we'll get the following result:

 

This graph provides additional insights into the keyword combinations that may be relevant for our content strategy. For instance, many people search for the keyword clusters for the gambling niche, the word "grouping" is used a lot (can be a synonym to target), also AI tools for keyword clustering are getting traction.

Morphological vs SERP-Based Clustering vs Knowledge Graphs

Most keyword clustering tools use one of two approaches: morphological or SERP-based clustering. Knowledge graph clustering combines the strengths of the both approaches and adds the powerful text network analysis method on top that helps rank the keywords by their relative influence and cluster them based on their co-occurrence in multiple contexts.

Morphological clustering looks at the words themselves — their roots, stems, and how they're structured linguistically. If you have "running shoes," "runners," and "run fast," a morphological tool groups them because they share the root "run." It's quick and doesn't cost anything since you're just analyzing text patterns, but it completely ignores what people are actually searching for. You might end up clustering "apple pie recipe" with "apple computer" just because both contain "apple," even though the search intent couldn't be more different.

SERP-based clustering takes a different approach by actually querying search engines and comparing the top 10 results. If "best espresso machine" and "top espresso makers" show 6 out of 10 identical URLs in their results, the tool assumes Google sees them as having the same intent and clusters them together. This method is slower and costs money (you're hitting APIs for every keyword), but it reflects how search engines actually interpret user intent. You can adjust the clustering threshold — set it to 30% overlap for broader clusters or 70% for tighter grouping. The downside is that you're always chasing Google's current rankings, which shift over time, and you're limited by API rate limits when working with large keyword lists.

Content Strategy: Topical Authority based on Keyword Clustering

Once you've identified your authority opportunities through keyword clustering, you can translate them into a concrete content production strategy that will help you establish topical authority. Here's a step-by-step workflow:

1. Map Clusters to Content Types

Based on your keyword cluster analysis, assign content types that match search intent and authority needs first. The task here is to create the pillar guide page that has a high concentration of all the different keyword clusters and then separately publish the pages that cover the content gaps as well as the specific topics identified.

Cluster Characteristics	Content Type	Authority Signal
High density, central position	Pillar Guide (3000+ words)	Comprehensive coverage of all clusters for topical authority
Gap between search intent & results	Unique Angle Article (1500 words)	Novel connections between clustersfor thought leadership
High betweenness keywords	Comparison/vs. Content	Connects topics for information hub
Peripheral but growing clusters	Tutorial/How-to (1000 words)	Early coverage of long-tail keywords for emerging authority
Frequently co-occurring pairs	FAQ/Q&A Page	Answers related queries for entity relationship and AI optimization (people use questions to find information in chatbots)

 

2. Publish the Content

Now that the content types are identified, you can publish the content starting from the pillar guide page and then the unique angle article, comparison/vs. content, tutorial/how-to, and FAQ/Q&A page.

1. Foundation Pillar
   Publish your main pillar page covering the highest-centrality cluster. Even if spoke pages don't exist yet, link to their planned URLs with placeholder anchor text from your graph.
2. Essential Spokes
   Create 3-5 spoke pages covering high-betweenness keywords that connect to your pillar. Each should link back to pillar AND to each other where keyword co-occurrence data shows relationships.
3. Gap Opportunities
   Write content targeting the search intent gaps you discovered. These should still link to your pillar, positioning you as covering aspects competitors miss.
4. Supporting Content
   Add FAQ pages, comparison content, and tutorials that round out the cluster and catch long-tail variations.

 

3. Graph-Informed Link Architecture

Use the visual structure of your keyword graph as a blueprint for internal linking. The idea is to link the pages that are related to each other based on the keyword co-occurrence in the graph.

• Node connections = page links: If two keywords appear together frequently in the graph, the pages targeting those keywords should link to each other
• Edge weight = link prominence: Stronger connections (thicker edges) in your graph should use more prominent contextual links (in intro paragraphs) between pages
• Cluster density = linking density: Tightly clustered topics should have more internal links per page (aim for 5-8 contextual links)
• Hub nodes = hub pages: Pages targeting high-betweenness keywords should be your most-linked pages (both to and from)

 

4. Measure Authority Gains

Re-analyze your keyword clusters monthly to measure authority growth. The key indicators are the following:

• Cluster ranking improvements: Are you now appearing in top 10 for entire keyword clusters vs. individual keywords?
• SERP feature captures: Are your pages winning featured snippets, AI-generated search results, People Also Ask, or related searches?
• Entity recognition: When you import current SERP content for your main query, does your content appear in the knowledge graph?
• Gap closure rate: Are the gaps you targeted now showing your content in search results?

 

5. Content Refresh Strategy

As search intent evolves, your authority content must too. Use InfraNodus quarterly to:

• Re-analyze top search results for your main keywords
• Compare current graph to your original authority content
• Identify new clusters emerging (new topics gaining traction)
• Identify dying clusters (topics becoming less relevant)
• Update pillar pages to incorporate new clusters
• Archive or redirect content for dying clusters

 

Automated Keyword Clustering with AI: InfraNodus MCP

For developers, researchers, and SEO professionals looking to automate keyword clustering workflows, InfraNodus offers a Model Context Protocol (MCP) integration that brings AI-powered keyword clustering directly into your development environment or AI applications. This enables automated keyword clustering, search intent analysis, content gap detection, and topical authority building without manual intervention. You can use it in your Python scripts, IDEs, or Claude Code / Claude chat applications.

What is InfraNodus MCP?

The InfraNodus MCP is a standardized protocol that connects large language models (LLMs) like Claude, ChatGPT, and other AI systems directly to InfraNodus's knowledge graph analysis engine via its API. This means you can perform keyword clustering and SEO analysis using natural language commands in your AI chat interface, code editor, or automated workflows.

Key Features for Keyword Clustering

The InfraNodus MCP server has a collection of tools that can be called directly from your LLM workflow. Some of the use cases include:
 

• Automated Search Intent Analysis: Generate keyword clusters from Google search queries, related searches, and People Also Ask data automatically
• SERP-Based Clustering: Pull and analyze top search results to identify keyword co-occurrence patterns and topical clusters
• Content Gap Detection: Compare search intent graphs with search results graphs to identify missing keyword combinations
• Topical Authority Mapping: Generate knowledge graphs showing relationships between keyword clusters and content topics
• AI-Generated Insights: Get research questions, content ideas, and SEO recommendations based on the keyword clustering analysis
• Python Integration: Build custom keyword clustering pipelines and SEO automation workflows using the MCP protocol

Example: Automated Keyword Clustering Workflow

You can use InfraNodus MCP with an AI assistant like Claude or via inside Cursor IDE to improve SEO for a web page. Just open the page in the IDE and provide the following prompt:

The example above will give you (and the model) a good idea of the main topics and gaps inside. As the next step, you can ask it to analyze the current search intent for the top 5 keyword combinations:

 

Try InfraNodus Keyword Clustering Yourself

You can try keyword clustering using the InfraNodus templates for your own use cases:

Log In    Sign Up