Introducing SEOntology: The Future Of SEO In The Age Of AI via @sejournal, @cyberandy

Screenshot from WebVowl, August 2024SEOntology – a snapshot (see an interactive visualization here).

The Data Interoperability Challenge In The SEO Industry

Let’s commencement tiny and reappraisal the benefits of a shared ontology with a applicable illustration (here is simply a descent taken from Emilija Gjorgjevska’s presentation astatine this year’s ZagrebSEOSummit)

Image from Emilija Gjorgjevska’s, ZagrebSEOSummit, August 2024

Imagine your workfellow Valentina uses a Chrome hold to export information from Google Search Console (GSC) into Google Sheets. The information includes columns similar “ID,” “Query,” and “Impressions” (as shown connected the left). But Valentina collaborates with Jan, who’s gathering a concern furniture utilizing the aforesaid GSC data. Here’s the problem: Jan uses a antithetic naming normal (“UID,” “Name,” “Impressionen,” and “Klicks”).

Now, standard this script up. Imagine moving with n antithetic information partners, tools, and squad members, each utilizing assorted languages. The effort to perpetually construe and reconcile these antithetic naming conventions becomes a large obstacle to effectual information collaboration.

Significant worth gets mislaid successful conscionable trying to marque everything enactment together. This is wherever an SEO ontology comes in. It is simply a communal language, providing a shared sanction for the aforesaid conception crossed antithetic tools, partners, and languages.

By eliminating the request for changeless translation and reconciliation, an SEO ontology streamlines information collaboration and unlocks the existent worth of your data.

The Genesis Of SEOntology

In the past year, we person witnessed the proliferation of AI Agents and the wide adoption of Retrieval Augmented Generation (RAG) successful each its antithetic forms (Modular, Graph RAG, and truthful on).

RAG represents an important leap guardant successful AI technology, addressing a cardinal regulation of traditional ample connection models (LLMs) by letting them entree outer knowledge.

Traditionally, LLMs are similar libraries with 1 publication – constricted by their grooming data. RAG unlocks a immense web of resources, allowing LLMs to supply much broad and close responses.

RAGs amended factual accuracy, and discourse understanding, perchance reducing bias. While promising, RAG faces challenges successful information security, accuracy, scalability, and integration, especially successful the endeavor sector.

For palmy implementation, RAG requires high-quality, structured data that tin beryllium easy accessed and scaled.

We’ve been among the archetypal to experimentation with AI Agents and RAG powered by the Knowledge Graph successful the discourse of contented instauration and SEO automation.

Screenshot from Agent WordLift, August 2023

Knowledge Graphs (KGs) Are Indeed Gaining Momentum In RAG Development

Microsoft’s GraphRAG and solutions similar LlamaIndex show this. Baseline RAG struggles to link accusation crossed disparate sources, hindering tasks requiring a holistic knowing of ample datasets.

KG-powered RAG approaches similar the 1 offered by LlamaIndex successful conjunction with WordLift code this by creating a cognition graph from website information and utilizing it alongside the LLM to amended effect accuracy, peculiarly for analyzable questions.

Image from author, August 2024

We person tested workflows with clients successful antithetic verticals for implicit a year.

From keyword probe for ample editorial teams to the procreation of question and answers for ecommerce websites, from contented bucketing to drafting the outline of a newsletter oregon revamping existing articles, we’ve been investigating antithetic strategies and learned a fewer things on the way:

1. RAG Is Overhyped

It is simply 1 of galore improvement patterns that execute a extremity of higher complexity. A RAG (or Graph RAG) is meant to assistance you prevention clip uncovering an answer. It’s superb but doesn’t lick immoderate selling tasks a squad indispensable grip daily. You request to absorption connected the information and the information model.

While determination are bully RAGs and atrocious RAGs, the cardinal differentiation is often represented by the “R” portion of the equation: the Retrieval. Primarily, the retrieval differentiates a fancy demo from a real-world application, and down a bully RAG, determination is ever bully data. Data, though, is not conscionable immoderate benignant of information (or graph data).

It is built astir a coherent information exemplary that makes consciousness for your usage case. If you physique a hunt motor for wines, you request to get the champion dataset and exemplary the information astir the features a idiosyncratic volition trust connected erstwhile looking for information.

So, information is important, but the information exemplary is adjacent much important. If you are gathering an AI Agent that has to bash things successful your selling ecosystem, you indispensable exemplary the information accordingly. You privation to correspond the essence of web pages and contented assets.

Image from author, August 2024

2. Not Everyone Is Great At Prompting

Expressing a task successful written signifier is hard. Prompt engineering is going astatine afloat velocity towards automation (here is my nonfiction connected going from prompting to punctual programming for SEO) arsenic lone a fewer experts tin constitute the punctual that brings america to the expected outcome.

This poses respective challenges for the plan of the idiosyncratic acquisition of autonomous agents. Jakon Nielsen has been precise vocal astir the antagonistic interaction of prompting connected the usability of AI applications:

“One large usability downside is that users indispensable beryllium highly articulate to constitute the required prose substance for the prompts.”

Even successful affluent Western countries, statistic provided by Nielsen archer america that only 10% of the colonisation tin afloat utilize AI! 

3. You Shall Build Workflows To Guide The User

The acquisition learned is that we indispensable physique elaborate standard operating procedures (SOP) and written protocols that outline the steps and processes to guarantee consistency, quality, and ratio successful executing peculiar optimization tasks.

We tin spot empirical grounds of the emergence of punctual libraries similar the 1 offered to users of Anthropic models oregon the unthinkable occurrence of projects similar AIPRM.

In reality, we learned that what creates concern worth is simply a bid of ci steps that assistance the idiosyncratic construe the discourse he/she is navigating successful into a accordant task definition.

We tin commencement to envision selling tasks similar conducting keyword probe arsenic a Standard Operating Procedure that tin usher the idiosyncratic crossed aggregate steps (here is however we mean the SOP for keyword find utilizing Agent WordLift)

4. The Great Shift To Just-in-Time UX 

In accepted UX design, accusation is pre-determined and tin beryllium organized successful hierarchies, taxonomies, and pre-defined UI patterns. As AI becomes the interface to the analyzable satellite of information, we’re witnessing a paradigm shift.

UI topologies thin to disappear, and the enactment betwixt humans and AI remains predominantly dialogic. Just-in-time assisted workflows tin assistance the idiosyncratic contextualize and amended a workflow.

• You request to think successful presumption of concern worth creation, absorption connected the user’s interactive journey, and facilitate the enactment by creating a UX connected the fly. Taxonomies stay a strategical asset, but they run down the scenes arsenic the idiosyncratic is teleported from 1 task to another, arsenic precocious brilliantly described by Yannis Paniaras from Microsoft.

Image from “The Shift to Just-In-Time UX: How AI is Reshaping User Experiences” by Yannis Paniaras, August 2024

5. From Agents To RAG (And GraphRAG) To Reporting

Because the idiosyncratic needs a concern interaction and RAG is lone portion of the solution, the absorption rapidly shifts from much generic questions and answering idiosyncratic patterns to precocious multi-step workflows.

The biggest issue, though, is what result the idiosyncratic needs. If we summation the complexity to seizure the highest concern goals, it is not capable to, let’s say, “query your data” oregon “chat with your website.”

A lawsuit wants a report, for example, of what is the thematic consistency of contented wrong the full website (this is simply a conception that we precocious discovered arsenic SiteRadus successful Google’s monolithic information leak), the overview of the seasonal trends crossed hundreds of paid campaigns, oregon the eventual reappraisal of the optimization opportunities related to the optimization of Google Merchant Feed.

You indispensable recognize however the concern operates and what deliverables you volition wage for. What factual actions could boost the business? What questions request to beryllium answered?

This is the commencement of creating a tremendous AI-assisted reporting tool.

How Can A Knowledge Graph (KG) Be Coupled With An Ontology For AI Alignment, Long-term Memory, And Content Validation?

The 3 guiding principles down SEOntology:

• Making SEO information interoperable to facilitate the instauration of cognition graphs portion reducing unneeded crawls and vendor locked-in;
• Infusing SEO know-how into AI agents utilizing a domain-specific language.
• Collaboratively sharing cognition and tactics to amended findability and prevent misuse of Generative AI.

When you woody with astatine slightest 2 information sources successful your SEO automation task, you volition already spot the vantage of utilizing SEOntology.

SEOntology As “The USB-C Of SEO/Crawling Data”

Standardizing data about contented assets, products, idiosyncratic hunt behavior, and SEO insights is strategic. The extremity is to person a “shared representation” of the Web arsenic a connection channel.

Let’s instrumentality a measurement backward. How does a Search Engine correspond a web page? This is our starting constituent here. Can we standardize however a crawler would correspond information extracted from a website? What are the advantages of adopting standards?

Practical Use Cases

Integration With Botify And Dynamic Internal Linking

Over the past fewer months, we’ve been moving intimately with the Botify squad to make thing exciting: a Knowledge Graph powered by Botify’s crawl information and enhanced by SEOntology. This collaboration is opening up caller possibilities for SEO automation and optimization.

Leveraging Existing Data With SEOntology

Here’s the chill part: If you’re already utilizing Botify, we tin pat into that goldmine of information you’ve collected. No request for further crawls oregon other enactment connected your part. We usage the Botify Query Language (BQL) to extract and alteration the needed information utilizing SEOntology.

Think of SEOntology arsenic a cosmopolitan translator for SEO data. It takes the analyzable accusation from Botify and turns it into a format that’s not conscionable machine-readable but machine-understandable. This allows america to make a rich, interconnected Knowledge Graph filled with invaluable SEO insights.

What This Means for You

Once we person this Knowledge Graph, we tin bash immoderate beauteous astonishing things:

• Automated Structured Data: We tin automatically make structured information markup for your merchandise listing pages (PLPs). This helps hunt engines amended recognize your content, perchance improving your visibility successful hunt results.
• Dynamic Internal Linking: This is wherever things get truly interesting. We usage the information successful the Knowledge Graph to make smart, dynamic interior links crossed your site. Let maine interruption down however this works and wherefore it’s truthful powerful.

In the diagram below, we tin besides spot however information from Botify tin beryllium blended with information from Google Search Console.

While successful astir implementations, Botify already imports this information into its crawl projects, erstwhile this is not the case, we tin trigger a caller API petition and import clicks, impressions, and positions from GSC into the graph.

Collaboration With Advertools For Data Interoperability

Similarly, we collaborated with the superb Elias Dabbas, creator of Advertools — a favourite Python room among marketers – to automate a wide scope of selling tasks.

Our associated efforts purpose to heighten information interoperability, allowing for seamless integration and information speech crossed antithetic platforms and tools.

In the first Notebook, disposable successful the SEOntology GitHub repository, Elias showcases however we tin effortlessly conception attributes for the WebPage class, including title, meta description, images, and links. This instauration enables america to easy exemplary analyzable elements, specified arsenic interior linking strategies. See present the structure:

• Internal_Links

• • anchorTextContent
  • NoFollow
  • Link

We tin besides adhd a emblem if the leafage is already utilizing schema markup:

• usesSchema

Formalizing What We Learned From The Analysis Of The Leaked Google Search Documents

While we privation to beryllium highly conscious successful deriving tactics oregon tiny schemes from Google’s monolithic leak, and we are good alert that Google volition rapidly forestall immoderate imaginable misuse of specified information, determination is simply a large level of accusation that, based connected what we learned, tin beryllium utilized to amended however we correspond web contented and signifier selling data.

Despite these constraints, the leak offers invaluable insights into improving web contented practice and selling information organization. To democratize entree to these insights, I’ve developed a Google Leak Reporting tool designed to marque this accusation readily disposable to SEO pros and integer marketers.

For instance, knowing Google’s classification strategy and its segmentation of websites into assorted taxonomies has been peculiarly enlightening. These taxonomies – specified arsenic ‘verticals4’, ‘geo’, and ‘products_services’ – play a important relation successful hunt ranking and relevance, each with unsocial attributes that power however websites and contented are perceived and ranked successful hunt results.

By leveraging SEOntology, we tin follow immoderate of these attributes to heighten website representation.

Now, intermission for a 2nd and ideate transforming the analyzable SEO information you negociate regular done tools similar Moz, Ahrefs, Screaming Frog, Semrush, and galore others into an interactive graph. Now, envision an Autonomous AI Agent, specified arsenic Agent WordLift, astatine your side.

This cause employs neuro-symbolic AI, a cutting-edge attack that combines neural learning capabilities with symbolic reasoning, to automate SEO tasks similar creating and updating interior links. This streamlines your workflow and introduces a level of precision and ratio antecedently unattainable.

SEOntology serves arsenic the backbone for this vision, providing a structured model that enables the seamless speech and reuse of SEO information crossed antithetic platforms and tools. By standardizing however SEO information is represented and interconnected, SEOntology ensures that invaluable insights derived from 1 instrumentality tin beryllium easy applied and leveraged by others. For instance, information connected keyword show from SEMrush could pass contented optimization strategies successful WordLift, each wrong a unified, interoperable environment. This not lone maximizes the inferior of existing information but besides accelerates the automation and optimization processes that are important for effectual marketing.

Infusing SEO Know-How Into AI Agents

As we make a caller agentic attack to SEO and integer marketing, SEOntology serves arsenic our domain-specific connection (DSL) for encoding SEO skills into AI agents. Let’s look astatine a applicable illustration of however this works.

Screenshot from WordLift, August 2024

We’ve developed a strategy that makes AI agents alert of a website’s integrated hunt performance, enabling a caller benignant of enactment betwixt SEO professionals and AI. Here’s however the prototype works:

System Components

• Knowledge Graph: Stores Google Search Console (GSC) data, encoded with SEOntology.
• LLM: Translates earthy connection queries into GraphQL and analyzes data.
• AI Agent: Provides insights based connected the analyzed data.

Human-Agent Interaction

Image from author, August 2024

The diagram illustrates the travel of a emblematic interaction. Here’s what makes this attack powerful:

• Natural Language Interface: SEO professionals tin inquire questions successful plain connection without constructing analyzable queries.
• Contextual Understanding: The LLM understands SEO concepts, allowing for much nuanced queries and responses.
• Insightful Analysis: The AI cause doesn’t conscionable retrieve data; it provides actionable insights, specified as:
  • Identifying top-performing keywords.
  • Highlighting important show changes.
  • Suggesting optimization opportunities.
• Interactive Exploration: Users tin inquire follow-up questions, enabling a dynamic exploration of SEO performance.

By encoding SEO cognition done SEOntology and integrating show data, we’re creating AI agents that tin supply context-aware, nuanced assistance successful SEO tasks. This attack bridges the spread betwixt earthy information and actionable insights, making precocious SEO investigation much accessible to professionals astatine each levels.

This illustration illustrates however an ontology similar SEOntology tin empower america to physique agentic SEO tools that automate analyzable tasks portion maintaining quality oversight and ensuring prime outcomes. It’s a glimpse into the aboriginal of SEO, wherever AI augments quality expertise alternatively than replacing it.

Human-In-The-Loop (HTIL) And Collaborative Knowledge Sharing

Let’s beryllium crystal clear: While AI is revolutionizing SEO and Search, humans are the beating bosom of our industry. As we dive deeper into the satellite of SEOntology and AI-assisted workflows, it’s important to recognize that Human-in-the-Loop (HITL) isn’t conscionable a fancy add-on—it’s the instauration of everything we’re building.

The essence of creating SEOntology is to transportation our corporate SEO expertise to machines portion ensuring we, arsenic humans, stay firmly successful the driver’s seat. It’s not astir handing implicit the keys to AI; it’s astir teaching it to beryllium the eventual co-pilot successful our SEO journey.

Human-Led AI: The Irreplaceable Human Element

SEOntology is much than a method model – it’s a catalyst for collaborative cognition sharing that emphasizes quality imaginable successful SEO. Our committedness extends beyond codification and algorithms to nurturing skills and expanding the capabilities of new-gen marketers and SEO pros.

Why? Because AI’s existent powerfulness successful SEO is unlocked by quality insight, divers perspectives, and real-world experience. After years of moving with AI workflows, I’ve realized that agentive SEO is fundamentally human-centric. We’re not replacing expertise; we’re amplifying it.

We present much businesslike and trustworthy results by blending cutting-edge tech with quality creativity, intuition, and ethical judgment. This attack builds spot with clients wrong our manufacture and crossed the web.

Here’s wherever humans stay irreplaceable:

• Understanding Business Needs: AI tin crunch numbers but can’t regenerate the nuanced knowing of concern objectives that seasoned SEO professionals bring. We request experts who tin construe lawsuit goals into actionable SEO strategies.
• Identifying Client Constraints: Every concern is unique, with its limitations and opportunities. It takes quality penetration to navigate these constraints and make tailored SEO approaches that enactment wrong real-world parameters.
• Developing Cutting-Edge Algorithms: The algorithms powering our AI tools don’t materialize retired of bladed air. We request superb minds to make state-of-the-art algorithms, larn from quality input, and continually improve.
• Engineering Robust Systems: Behind each smooth-running AI instrumentality is simply a squad of bundle engineers who guarantee our systems are fast, secure, and reliable. This quality expertise keeps our AI assistants moving similar well-oiled machines.
• Passion for a Better Web: At the bosom of SEO is simply a committedness to making the web a amended place. We request radical who stock Tim Berners’s—Lee’s vision—people who are passionate astir processing the web of information and improving the integer ecosystem for everyone.
• Community Alignment and Resilience: We request to unite to analyse the behaviour of hunt giants and make resilient strategies. It’s astir solving our problems innovatively arsenic individuals and arsenic a corporate force. This is what I ever loved astir the SEO industry!

Extending The Reach Of SEOntology

As we proceed to make SEOntology, we’re not operating successful isolation. Instead, we’re gathering upon and extending existing standards, peculiarly Schema.org, and pursuing the palmy exemplary of the GS1 Web Vocabulary.

SEOntology As An Extension Of Schema.org

Schema.org has go the de facto modular for structured information connected the web, providing a shared vocabulary that webmasters tin usage to markup their pages.

However, portion Schema.org covers a wide scope of concepts, it doesn’t delve profoundly into SEO-specific elements. This is wherever SEOntology comes in.

An hold of Schema.org, similar SEOntology, is fundamentally a complementary vocabulary that adds caller types, properties, and relationships to the halfway Schema.org vocabulary.

This allows america to support compatibility with existing Schema.org implementations portion introducing SEO-specific concepts not covered successful the halfway vocabulary.

Learning From GS1 Web Vocabulary

The GS1 Web Vocabulary offers a large exemplary for creating a palmy hold that interacts seamlessly with Schema.org. GS1, a planetary enactment that develops and maintains proviso concatenation standards, created its Web Vocabulary to widen Schema.org for e-commerce and merchandise accusation usage cases.

The GS1 Web Vocabulary demonstrates, adjacent recently, however industry-specific extensions tin power and interact with schema markup:

• Real-world impact: The https://schema.org/Certification property, present officially embraced by Google, originated from GS1’s https://www.gs1.org/voc/CertificationDetails. This showcases however extensions tin thrust the improvement of Schema.org and hunt motor capabilities.

We privation to travel a akin attack to widen Schema.org and go the modular vocabulary for SEO-related applications, perchance influencing aboriginal hunt motor capabilities, AI-driven workflows, and SEO practices.

Much similar GS1 defined their namespace (gs1:) portion referencing schema terms, we person defined our namespace (seovoc:) and are integrating the classes wrong the Schema.org hierarchy erstwhile possible.

The Future Of SEOntology

SEOntology is much than conscionable a theoretical framework; it’s a applicable instrumentality designed to empower SEO professionals and instrumentality makers successful an progressively AI-driven ecosystem.

Here’s however you tin prosecute with and payment from SEOntology.

If you’re processing SEO tools:

• Data Interoperability: Implement SEOntology to export and import information successful a standardized format. This ensures your tools tin easy interact with different SEOntology-compliant systems.
• AI-Ready Data: By structuring your information according to SEOntology, you’re making it much accessible for AI-driven automations and analyses.

If you’re an SEO professional:

• Contribute to Development: Just similar with Schema.org, you tin lend to SEOntology’s evolution. Visit its GitHub repository to:
  • Raise issues for caller concepts oregon properties you deliberation should beryllium included.
  • Propose changes to existing definitions.
  • Participate successful discussions astir the aboriginal absorption of SEOntology.
• Implement successful Your Work: Start utilizing SEOntology concepts successful your structured data.

In Open Source We Trust

SEOntology is an open-source effort, pursuing successful the footsteps of palmy projects similar Schema.org and different shared linked vocabularies.

All discussions and decisions volition beryllium public, ensuring the assemblage has a accidental successful SEOntology’s direction. As we summation traction, we’ll found a committee to steer its improvement and stock regular updates.

Conclusion And Future Work

The aboriginal of selling is human-led, not AI-replaced. SEOntology isn’t conscionable different buzzword – it’s a measurement towards this future. SEO is strategical for the improvement of agentive selling practices.

SEO is nary longer astir rankings; it’s astir creating intelligent, adaptive contented and fruitful dialogues with our stakeholders crossed assorted channels. Standardizing SEO information and practices is strategical to physique a sustainable aboriginal and to invest successful liable AI.

Are you acceptable to articulation this revolution?

There are 3 guiding principles down the enactment of SEOntology that we request to marque wide to the reader:

• As AI needs semantic data, we request to marque SEO information interoperable, facilitating the instauration of cognition graphs for everyone. SEOntology is the USB-C of SEO/crawling data. Standardizing information astir contented assets and products and however radical find content, products, and accusation successful wide is important. This is the archetypal objective. Here, we person 2 applicable usage cases. We person a connector for WordLift that gets crawl information from the Botify crawler and helps you jump-start a KG that uses SEOntology arsenic a information model. We are besides moving with Advertools, an open-source crawler and SEO tool, to marque information interoperable with SEOntology;
• As we advancement with the improvement of a caller agentic mode of doing SEO and integer marketing, we privation to infuse the know-how of SEO utilizing SEOntology, a domain-specific connection to infuse the SEO mindset to SEO agents (or multi-agent systems similar Agent WordLift). In this context, the accomplishment required to make dynamic interior links is encoded arsenic nodes successful a cognition graph, and opportunities go triggers to activate workflows.
• We expect to enactment with human-in-the-loop HITL, meaning that the ontology volition go a mode to collaboratively stock cognition and tactics that assistance amended findability and forestall the misuse of Generative AI that is polluting the Web today.

Project Overview

• SEOntology – Project Overview – WordLift
• Public repository
• Thread connected X
• R&D Work
• 6125 K-LAM – Enhancing Large Action Models with Knowledge Graphs.docx

More resources: 

• AI Has Changed How Search Works
• Leveraging Generative AI Tools For SEO
• How Search Engines Work

Featured Image: tech_BG/Shutterstock