Semantic SEO Encyclopedia

265 terms in 15 categories

AI Search

50-Word Rule

50-Word Rule — Content optimization rule for AI Search: key answers must appear within the first 50 words of an article or H2 section.

AI Search

Agent Decision Optimization

Agent Decision Optimization optimizes content for the moment AI agents decide whether to cite a source, not just for search rankings.

AI Search

AI Mode / AI Overview

AI Mode and AI Overview are Google features that display synthetic AI answers directly in search results above organic listings.

AI Search

AI SEO Alignment Score

AI SEO Alignment Score measures how well content matches AI Search citability criteria: factors in information density, BLUF, and SRL.

AI Search

API-able Brand

API-able Brand is a brand optimized for consistent retrieval by APIs and AI systems through unified naming, structured data, and presence.

AI Search

BLUF (Bottom Line Up Front)

BLUF (Bottom Line Up Front) is placing the key answer at the very beginning of content, increasing citation probability to 62%.

AI Search

ChatGPT (Bing + Broad Fanout)

ChatGPT (Bing + Broad Fanout) is OpenAI's AI Search platform using Bing's index with broad query fanout: prefers Wikipedia content.

AI Search

Cited (Being Cited)

Cited: Second visibility level in AI Search where content gets cited as a source in AI responses with a link or brand name provided.

AI Search

Claude (Own Index)

Claude (Own Index): Anthropic's AI platform with its own web index. Unlike ChatGPT, it doesn't rely on Bing for search results.

AI Search

Competitive Heuristic

Competitive Heuristic — The most effective AI Search content optimization heuristic (+1.61 weight), based on competitor comparisons (X vs Y).

AI Search

Content Freshness

Content Freshness measures how recently content was updated: AI Search favors current data and regular updates over outdated content.

AI Search

Cross-Domain Web Entities

Cross-Domain Web Entities are unified brand representations that AI systems recognize across multiple platforms and domains.

AI Search

Earned Media vs Owned Media

Earned Media vs Owned Media distinguishes content on your own channels (owned) from mentions in external sources (earned).

AI Search

Featured Snippet

Featured Snippet is a highlighted answer displayed at position zero in Google results, above organic listings—tables and lists have the highest chances.

AI Search

Framework Retrieved-Cited-Trusted

Framework Retrieved-Cited-Trusted is a three-stage AI Search visibility framework: being Retrieved from index, being Cited, and becoming Trusted.

AI Search

Gemini (Knowledge Graph + YouTube)

Gemini is Google's AI Search platform leveraging Google Knowledge Graph and YouTube for unique entity understanding and multimodal capabilities.

AI Search

Intentional Decomposition

Intentional Decomposition — A query breakdown method that maps user decision-making stages to anticipate follow-up questions throughout the customer jou...

AI Search

Keyword Stuffing (in AI Context)

Keyword Stuffing (in AI Context) is the SEO technique of mechanically repeating keywords, which becomes ineffective against AI-powered search systems.

AI Search

Linguistic Hedging

Linguistic Hedging is using words that weaken statement certainty (maybe, perhaps, probably), reducing likelihood of AI citations.

AI Search

Minimalist Heuristic

Minimalist Heuristic: AI Search optimization strategy with negative impact (weight -1.66 in studies); reducing content depth decreases AI citation oppor...

AI Search

Multisourcing

Multisourcing is establishing brand presence across multiple independent platforms so AI systems can cross-verify information and consider it credible.

AI Search

Perplexity (Aggressive Fanout)

Perplexity is an AI search platform that generates multiple related queries and always shows source links with real traffic opportunities.

AI Search

Query Fan-out

Query Fan-out is an AI Search mechanism that splits a single user query into 5–10 sub-queries, each searching the index independently.

AI Search

Reasoning Gap (Question Coverage Matrix)

Reasoning Gap is a question coverage matrix identifying AI reasoning gaps: questions no content on the web answers well despite the need.

AI Search

Retrieved (Being Retrieved from the Index)

Retrieved (Being Retrieved from the Index) is content that has been pulled from search indexes into an AI system's context window.

AI Search

Semantic Decomposition

Semantic Decomposition is a type of query decomposition in AI Search that breaks questions into component parts by meaning — e.g.

AI Search

Sub-queries

Sub-queries are smaller helper questions automatically generated by AI search systems, breaking complex queries into manageable parts.

AI Search

Thin Content (AI-generated)

Thin Content (AI-generated) is low-quality content mass-produced by AI without unique value: detected by algorithms and lowering rankings.

AI Search

Trusted (Being a Trusted Source)

Trusted (Being a Trusted Source) is the third visibility level in AI Search where a brand becomes AI's default trusted source for queries.

AI Search

Verification Decomposition

Verification Decomposition breaks complex queries into verifiable sub-questions to ensure answer accuracy across multiple sources in AI Search.

AI Search

Zero-Click Search

Zero-Click Search occurs when users get answers directly in search results or AI Search without clicking through to any source website.

Contextual Vector

Contextual Connections (links)

Contextual Connections are links between related content based on shared semantic attributes rather than matching keywords.

Contextual Vector

Contextual Hierarchy (H1-H2-H3)

Contextual Hierarchy (H1-H2-H3) is a heading structure that builds semantic context for AI Search — H1 defines the topic, H2 divides into chunks.

Contextual Vector

Contextual Structure (BLUF per H2)

Contextual Structure (BLUF per H2) is a principle where each H2 section starts with the key thesis in the first sentence: ready chunk for AI citation.

Contextual Vector

Contextual Vector is a contextual vector defining three optimization dimensions: heading hierarchy, BLUF structure, and internal linking.

E-E-A-T

Authority (autorytet)

Authority (autorytet): E-E-A-T component representing recognition of author/site as industry authority through credible backlinks and validation signals.

E-E-A-T

Cross-domain entity building

Cross-domain entity building is the practice of building entity recognition across multiple platforms simultaneously to strengthen E-E-A-T and Trust.

E-E-A-T

Experience

Experience is the E-E-A-T component representing the author's first-hand knowledge of a topic, demonstrated through case studies and original content.

E-E-A-T

Expertise

Expertise is an E-E-A-T component representing an author's professional knowledge, signaled by publications, certifications, and citations.

E-E-A-T

Trust

Trust is the credibility component of E-E-A-T that measures how reliable and secure a website appears to users and search engines.

E-E-A-T

TTS (Truthful Text Summarization)

TTS (Truthful Text Summarization) is a mechanism that verifies entity information consistency across sources—AI compares your data with network consensus.

Embeddings

Cannibalization (similarity 0.9–0.99)

Cannibalization (similarity 0.9–0.99) is when pages compete for the same queries, detected when semantic analysis shows 90-99% content similarity.

Embeddings

CLUSTERING

CLUSTERING is an embedding task type that optimizes vectors for topical grouping and is used in keyword clustering pipelines.

Embeddings

Content Pruning (outliers)

Content Pruning (outliers): identifying pages topically distant from a site's centroid (outliers) as candidates for removal or relocation.

Embeddings

Cosine Similarity (0–1)

Cosine Similarity (0–1) measures the angle between two vectors to determine content similarity — the standard metric for comparing embeddings in SEO.

Embeddings

Duplicate Detection (similarity 1.0)

Duplicate Detection (similarity 1.0) detects identical content using embeddings and cosine similarity near 1.0 for SEO duplicate identification.

Embeddings

Embedding Cache (historical)

Embedding Cache (historical) is a mechanism that stores computed embeddings in the data/embeddings directory to avoid repeated API calls.

Embeddings

Embedding Normalization

Embedding normalization scales embedding vectors to uniform length: 3072-dimensional embeddings are normalized 'out of the box'.

Embeddings

Euclidean Distance

Euclidean distance measures the straight-line distance between two points in vector space — smaller values indicate greater semantic similarity.

Embeddings

Generative Model

Generative Model is an AI that creates text from prompts (GPT-4, Claude, Gemini), unlike embedding models that convert text to vectors.

Embeddings

Internal Linking (nearest neighbors)

Internal Linking (nearest neighbors) is an internal linking strategy based on embeddings and the nearest neighbors algorithm.

Embeddings

MTEB Leaderboard

MTEB Leaderboard ranks embedding models on standardized benchmarks, helping SEO professionals choose the best embedding model for specific tasks.

Embeddings

Nearest Neighbors

Nearest Neighbors is an algorithm that finds the k closest points in vector space by measuring similarity between data points.

Embeddings

Redirect Maps (migration)

Redirect Maps (migration) are automated systems that use embeddings to match old URLs to new ones during site migration based on content similarity.

Embeddings

RETRIEVAL_DOCUMENT

RETRIEVAL_DOCUMENT is an embedding task type that optimizes vectors for representing documents — used on the document side in RAG systems.

Embeddings

RETRIEVAL_QUERY

RETRIEVAL_QUERY is an embedding task type that creates vectors optimized for search queries and is used on the query side in RAG systems.

Embeddings

Semantic Search Engine

Semantic Search Engine is a search system based on embeddings and cosine similarity that understands query intent rather than keywords.

Embeddings

SEMANTIC_SIMILARITY

SEMANTIC_SIMILARITY is an embedding task type optimizing vectors for measuring similarity between texts, used in duplicate detection and cannibalization.

Embeddings

t-SNE and PCA

t-SNE and PCA are dimensionality reduction methods that reduce high-dimensional vectors to 2D/3D for visualizing topical clusters.

Embeddings

Task Type (parameter)

Task Type is an embedding model parameter that specifies the intended task (retrieval, classification, clustering) to optimize vector quality.

Embeddings

Tokenization

Tokenization splits text into smaller units called tokens that AI models can process, such as words, subwords, or characters.

Embeddings

Transformer (architecture)

Transformer (architecture) is a neural network architecture created by Google that serves as the foundation for models like BERT and GPT.

Embeddings

UMAP (dimensionality reduction)

UMAP (dimensionality reduction) reduces embedding dimensions (e.g., from 768 to 5 dimensions) while preserving the most important relationships.

Embeddings

Vector Quantization

Vector quantization compresses embedding vectors by reducing numeric precision, trading slight accuracy loss for major memory savings.

Embeddings

Vector Representation

Vector Representation encodes information as numerical arrays, letting AI and search engines mathematically compare text meanings.

Embeddings

Word2Vec (2013)

Word2Vec (2013) is a pioneering embedding model that converts words to numeric vectors: learned from context but assigned each word one fixed vector.

Knowledge Graphs

Attribute Network (Semantic Hubs)

Attribute Network identifies attributes that connect multiple entities simultaneously, forming semantic hubs in graph analysis.

Knowledge Graphs

Betweenness Centrality (Hub Page)

Betweenness Centrality measures how many shortest paths between node pairs pass through a given node: identifying Hub Pages that bridge topical clusters.

Knowledge Graphs

Content Gaps (Graph vs Own Site)

Content Gaps (Graph vs Own Site) identifies missing content by comparing what the knowledge graph for a topic contains versus what your site covers.

Knowledge Graphs

Contextual Bridge

Contextual Bridge is a knowledge graph attribute linking two distant entities through shared context: e.g., 'taxation' bridging inheritance and donation.

Knowledge Graphs

Core Unique (Graph Layer)

Core Unique (Graph Layer) is the top tier of a knowledge graph containing unique and root attributes that form an entity's topical foundation.

Knowledge Graphs

Degree Centrality (Pillar Page)

Degree Centrality (Pillar Page) is a graph metric measuring node connections: high values indicate pillar page candidates.

Knowledge Graphs

EAV-to-Graph Mapping

EAV-to-Graph Mapping transforms Entity-Attribute-Value triplets into knowledge graph structure: Entity and Value become nodes.

Knowledge Graphs

Edge (Graph Edge)

Edge (Graph Edge) is a connection between two nodes in a knowledge graph representing a relationship—has a label (relationship type).

Knowledge Graphs

Graph-based vs Lexical Linking

Graph-based vs Lexical Linking compares two internal linking approaches: lexical (keyword matching) vs graph-based (semantic relationships).

Knowledge Graphs

GraphRAG (Microsoft)

GraphRAG (Microsoft) — Microsoft's approach combining knowledge graphs with RAG, searching graph structure instead of similar chunks.

Knowledge Graphs

Helicopter View

Helicopter View: A knowledge graph-level perspective on an entire site or topic that reveals complete topical structure at a glance.

Knowledge Graphs

HOP-PAA (Query Expansion Tag)

HOP-PAA is a query expansion tag marking sub-queries discovered by hopping through People Also Ask sections across multiple SERP levels.

Knowledge Graphs

Iterative Graph Expansion (MERGE)

Iterative Graph Expansion (MERGE) builds knowledge graphs by incrementally adding nodes and relationships using Neo4j's MERGE operation.

Knowledge Graphs

JSON (Graph Transport)

JSON (Graph Transport) is a pattern that enables structure transfer between Neo4j, Python, and LLMs using universal JSON format.

Knowledge Graphs

Knowledge Graph

Knowledge Graph is a data structure of nodes (entities) and edges (relationships) that represents knowledge about a topic.

Knowledge Graphs

LLM-PREDICTED (Query Expansion Tag)

LLM-PREDICTED marks sub-queries generated purely by language models without SERP validation — hypotheses requiring verification.

Knowledge Graphs

MERGE (Neo4j Operation)

MERGE (Neo4j Operation) creates nodes or relationships only if they don't already exist; otherwise, it updates their properties, preventing duplicates.

Knowledge Graphs

NODE (Details)

NODE (Details) — Node properties in knowledge graphs: type (entity/attribute), name, URR classification, layer (Core/Strong/Relevant).

Knowledge Graphs

Node (Graph Node)

Node (Graph Node) is a basic element of a knowledge graph representing an entity or attribute—a point from which edges connect to other nodes.

Knowledge Graphs

NODE Page

NODE Page is a site page that corresponds to a specific node in a knowledge graph — 1:1 mapping between nodes and URLs enables automatic.

Knowledge Graphs

Query Expansion (20–30 Sub-queries)

Query Expansion broadens a seed query into 20–30 sub-queries using LLM and SERP data: builds complete topical coverage of the central entity.

Knowledge Graphs

Relationship Label

Relationship Label names the connection between nodes in a knowledge graph, describing relationship types like HAS_ATTRIBUTE or SHARES_ATTRIBUTE.

Knowledge Graphs

Relationship Strength

Relationship Strength is the numeric value measuring connection strength between knowledge graph nodes — determining internal linking priority.

Knowledge Graphs

Relevant Contextual (Graph Layer)

Relevant Contextual is the lowest knowledge graph layer containing RARE attributes: contextual and supplementary information that supports main content.

Knowledge Graphs

SEED (Sub-topics)

SEED (Sub-topics) is a set of sub-topics derived from a seed query that forms the foundation for building topical clusters and knowledge graphs.

Knowledge Graphs

SEED Page

SEED Page is the starting page of a topical cluster corresponding to the central entity in the knowledge graph: the foundation for topical authority.

Knowledge Graphs

SEED-PAA (Query Expansion Tag)

SEED-PAA marks sub-queries from the People Also Ask box that appears for the seed query — The first and highest-priority expansion level.

Knowledge Graphs

SHARES_ATTRIBUTE (Shared Attributes)

SHARES_ATTRIBUTE is a knowledge graph relationship connecting entities that share the same attribute—determines internal linking strength between subpages.

Knowledge Graphs

Strong Direct (Graph Layer)

Strong Direct is the middle knowledge graph layer containing ROOT attributes with high search volumes: the ranking backbone content.

Lexical Semantics

Antonyms (Comparisons)

Antonyms (Comparisons) — Lexical relationship between words with opposite meanings that activates comparison frames and strengthens content.

Lexical Semantics

Boolean (Is X?)

Boolean (Is X?) is a semantic frame question type asking 'is X a Y?' that requires a yes/no answer with justification: often appears in PAA.

Lexical Semantics

Co-occurrences

Co-occurrences measure how frequently terms appear together in texts: strong co-occurrences build algorithmic expectations about content.

Lexical Semantics

Comparative (X vs Y?)

Comparative (X vs Y?) is a frame question type that requires comparison with criteria and shows high citation rates in AI search results.

Lexical Semantics

Cost (How Much?)

Cost (How Much?) is a semantic frame that addresses pricing questions requiring specific numerical data; AI Search favors exact amounts with context.

Lexical Semantics

Definitional (What is?)

Definitional (what is?) is a semantic frame question type asking 'what is X?' that requires an entity definition, typically covered at article start.

Lexical Semantics

Distributional Semantics

Distributional Semantics is a theory that word meaning comes from the contexts where words appear: the foundation of embeddings and semantic search.

Lexical Semantics

Frame Semantics

Frame Semantics is a theory that words activate holistic conceptual frames: 'purchase' triggers buyer, seller, price, and product elements.

Lexical Semantics

Grouping (What Types?)

Grouping is a semantic frame query type asking 'what are the types of X?' requiring taxonomy/classification; ideal format is lists or tables.

Lexical Semantics

Hypernyms (Superordinate Categories)

Hypernyms are broader category terms in conceptual hierarchies (e.g., 'vehicle' for 'car') that help search engines understand taxonomic relationships.

Lexical Semantics

Hyponyms (Entity Subtypes)

Hyponyms are entity subtypes in lexical relationships: 'espresso' is a hyponym of 'coffee'. They build topical depth and support Query Fan-out.

Lexical Semantics

Meronyms (Component Parts)

Meronyms (Component Parts): Lexical relationship indicating a component part of an entity (e.g., 'keyboard' is a meronym of 'laptop').

Lexical Semantics

Polysemy

Polysemy: when one word has multiple meanings (e.g., 'bank' = financial institution / riverbank) — an SEO challenge solved by contextual embeddings.

Lexical Semantics

Process (How to?)

Process (How to?) is a frame question type that answers 'how to do X?' with step-by-step instructions that lower Cost of Retrieval.

Lexical Semantics

Synonyms (Broader Matching)

Synonyms (Broader Matching) are words with the same or similar meaning — using synonyms in content broadens matching with user queries.

Lexical Semantics

Term Distribution in Articles

Term Distribution in Articles is the strategic placement of key terms across an article to maintain semantic relevance in all sections.

Macro-semantics (site level)

Breadcrumbs (Cluster Hierarchy)

Breadcrumbs (Cluster Hierarchy) show a page's hierarchical path within a site's cluster structure, helping Google understand site architecture.

Macro-semantics (site level)

Broad Core Update and Semantic Distance

Broad Core Update and Semantic Distance is the phenomenon where Google's Core Updates alter semantic relationships between topics, shifting clusters.

Macro-semantics (site level)

Content Consolidation

Content Consolidation merges semantically similar pages that cannibalize each other into one stronger, unified content piece.

Macro-semantics (site level)

Embedding Centroid

Embedding Centroid is the center of gravity of all page embedding vectors on a site: the point that defines 'what the site is about' in semantic space.

Macro-semantics (site level)

Faceted Navigation (Filters)

Faceted Navigation (Filters) is a filter system on e-commerce sites (color, size, price) generating dynamic URLs — requires careful configuration.

Macro-semantics (site level)

Google Warehouse API (Documentation Leak)

Google Warehouse API (Documentation Leak) — Internal Google documentation leak revealing metrics like Site Focus Score and Site Radius.

Macro-semantics (site level)

Navigation and Crawl Budget

Navigation and Crawl Budget describes how excessive navigation HTML consumes crawl budget, reducing resources available for content discovery.

Macro-semantics (site level)

Pillar Page

Pillar Page is the main page of a topical cluster covering the topic completely and linking to supporting pages; every CORE cluster needs one.

Macro-semantics (site level)

Server-Side Rendering (SSR)

Server-Side Rendering (SSR) generates HTML on the server instead of the browser, ensuring content is immediately available to search crawlers.

Macro-semantics (site level)

Site Architecture

Site Architecture is the hierarchical structure of a website (ROOT > SEED > NODE) that determines how crawlers and users navigate the site.

Macro-semantics (site level)

Site Focus Score

Site Focus Score is a metric from the Google Warehouse API leak measuring website topical coherence: higher focus means Google understands better.

Macro-semantics (site level)

Site Radius

Site Radius is a metric from the Google Warehouse API leak measuring a site's topical spread—small radius means focused, large means scattered.

Macro-semantics (site level)

Site-wide N-grams

Site-wide N-grams analyze the most frequent phrases across a website, revealing dominant topics and helping assess topical coherence.

Macro-semantics (site level)

TTFB (Time to First Byte)

TTFB (Time to First Byte) measures the time from an HTTP request to the first server response byte—a key metric for crawl budget and Core Web Vitals.

Macro-semantics (site level)

Two-wave Indexing

Two-wave Indexing is Google's two-phase indexing process: first wave analyzes raw HTML content, second wave processes JavaScript-rendered content.

Metrics & Audit

AI Citability Score (0-10)

AI Citability Score (0-10) measures how likely content is to be cited by AI systems — Evaluates chunk autonomy, BLUF, and atomic claims.

Metrics & Audit

AI Presence Rate

AI Presence Rate measures how often a brand appears in answers generated by AI Search systems — the AI-era equivalent of Share of Voice.

Metrics & Audit

BEFORE/AFTER (recommendations)

BEFORE/AFTER (recommendations) is an audit format that presents specific content changes with concrete before/after examples.

Metrics & Audit

Citation Authority

Citation Authority measures the quality of sources citing a brand in AI Search. More credible sources mean higher citation authority.

Metrics & Audit

Content Format Intelligence

Content Format Intelligence analyzes format preferences in top search results (FAQ, tables, lists) to complement traditional content gap analysis.

Metrics & Audit

CQS (Content Quality Score 0-100)

CQS (Content Quality Score 0-100) is a composite quality metric: a weighted average of six dimensions including CSI, CoR, Density, SRL, TF-IDF, E-E-A-T.

Metrics & Audit

CQS Formula (weighted average)

CQS Formula is a weighted scoring system combining six content quality dimensions: CSI (0.25), E-E-A-T (0.20), CoR (0.20), Density, SRL, and TF-IDF.

Metrics & Audit

Google Quality Rater Guidelines

Google Quality Rater Guidelines: Google's official document describing how to evaluate website quality, distinguishing Main Content.

Metrics & Audit

Impact × Effort (prioritization)

Impact × Effort prioritizes audit recommendations using Priority = Impact × (1/Effort), ranking from '1-NOW' (high impact, low effort) to '5-SKIP'.

Metrics & Audit

Schema.org Markup

Schema.org markup is structured data that uses the Schema.org vocabulary to help search engines understand content type and structure.

Metrics & Audit

Share of AI Conversation

Share of AI Conversation measures a brand's portion of AI-generated answers compared to competitors — the Share of Voice equivalent for AI Search.

Metrics & Audit

URR Classification (Unique/Root/Rare)

URR Classification categorizes entity attributes into three tiers: UNIQUE (differentiators), ROOT (definitions), and RARE (extras).

Micro-semantics (passage level)

Agent (Action Performer)

Agent (Action Performer): Semantic role (SRL) denoting the action performer in a sentence ('who does it'); essential for unambiguous quotable content.

Micro-semantics (passage level)

Atomic Claims (Verifiable)

Atomic Claims are indivisible, verifiable factual statements extracted from text: AI Search more easily cites fragments with Atomic Claims.

Micro-semantics (passage level)

Beneficiary (Recipient)

Beneficiary (Recipient): Semantic role (SRL) denoting who receives an action's benefit ('for whom it is'); helps AI personalize responses.

Micro-semantics (passage level)

Cost of Retrieval (CoR)

Cost of Retrieval (CoR): cost of extracting information from text fragments. The lower the better (BLUF, tables, lists, facts first).

Micro-semantics (passage level)

Entity Salience

Entity Salience measures an entity's semantic prominence in text, based on SRL role and sentence position within content structure.

Micro-semantics (passage level)

Fluff (Filler Words)

Fluff (filler words) are sentences with no informational value: generalities and rhetorical questions that lower Information Density.

Micro-semantics (passage level)

Information Density

Information Density measures the ratio of citable information to text volume, with higher concrete data per paragraph increasing density.

Micro-semantics (passage level)

Information Gain

Information Gain measures how much new, unique information a text fragment contributes compared to what already exists in the search engine's index.

Micro-semantics (passage level)

Instrument (Tool)

Instrument is a semantic role (SRL) denoting the tool used to perform an action — 'with what'; specifies context and increases citability.

Micro-semantics (passage level)

Location (Place)

Location (Place) is a semantic role denoting where an action takes place. Critical for local SEO and content precision in natural language.

Micro-semantics (passage level)

Main Content (MC)

Main Content (MC) is the part of a page that directly helps the page achieve its purpose, as defined by Google's Quality Rater Guidelines.

Micro-semantics (passage level)

Passage Embeddings

Passage Embeddings are vectors for individual page fragments, enabling Google to index and rank specific sections rather than whole documents.

Micro-semantics (passage level)

Passage Ready (seated ready)

Passage Ready is the state where every sentence in a text fragment is self-contained and ready for AI citation without additional context.

Micro-semantics (passage level)

Patient (Action Object)

Patient (Action Object) is the entity that receives or undergoes an action in semantic role labeling; precise identification reduces Cost of Retrieval.

Micro-semantics (passage level)

Semantic Role Labels (SRL)

Semantic Role Labels (SRL) assigns semantic roles: Agent (who), Predicate (what), Patient (what receives action), Beneficiary, Instrument, Location.

Micro-semantics (passage level)

Supplementary Content

Supplementary Content is additional page content (menus, links, ads, sidebars) that doesn't directly serve the page's main purpose.

RAG (Retrieval Augmented Generation)

Bi-encoder vs Cross-encoder

Bi-encoder vs Cross-encoder represents two distinct neural ranking approaches: fast separate vectors vs slow joint analysis for precision.

RAG (Retrieval Augmented Generation)

Chunk Autonomy

Chunk Autonomy is a content optimization principle ensuring every H2 section can be understood independently without context from other sections.

RAG (Retrieval Augmented Generation)

Chunking

Chunking splits documents into smaller fragments before vectorization in RAG systems, balancing semantic precision with context preservation.

RAG (Retrieval Augmented Generation)

Knowledge Cut-Off

Knowledge Cut-Off is the training data boundary date for an AI model: information after this date requires RAG or web search.

RAG (Retrieval Augmented Generation)

RAG (Retrieval Augmented Generation) combines information retrieval with AI text generation to provide fresh context on demand.

RAG (Retrieval Augmented Generation)

Reranking

Reranking is the second retrieval stage in RAG where a cross-encoder re-sorts initially retrieved results with greater precision than embeddings alone.

RAG (Retrieval Augmented Generation)

Supabase (Postgres + pgvector)

Supabase (Postgres + pgvector) is a database platform that stores and searches embeddings alongside traditional relational data in one service.

Semantic Audit Pipelines

Competitor Gap Analysis

Competitor gap analysis: A stage or pipeline comparing competitor content with your site at the semantic embedding level.

Semantic Audit Pipelines

Consolidated Markdown

Consolidated Markdown — a file combining multiple content sources into one consolidated Markdown document serving as input for LLM analysis.

Semantic Audit Pipelines

Content Brief

Content Brief — detailed article specification generated from pipeline data, containing goal, keywords, and H2/H3 structure.

Semantic Audit Pipelines

Content Format Recommendations

Content Format Recommendations are format suggestions (article, FAQ, list, infographic) based on SERP analysis and query intent.

Semantic Audit Pipelines

Graceful Degradation

Graceful Degradation is a pipeline design principle where individual step failures don't halt the entire process—it continues with reduced quality.

Semantic Audit Pipelines

Human in the Loop

Human in the Loop is a design pattern where humans verify and approve key decisions in AI-driven processes, balancing automation with quality control.

Semantic Audit Pipelines

Jina Reader (tool)

Jina Reader is a Jina AI tool that converts any web page to clean Markdown that strips HTML/CSS noise and produces text ready for AI analysis.

Semantic Audit Pipelines

LLM for Reasoning, Python for Computation

LLM for Reasoning, Python for Computation is a principle that divides tasks between language models for reasoning and Python code for computation.

Semantic Audit Pipelines

Noise Cleaning

Noise cleaning removes irrelevant data from SEO datasets by filtering out branded queries, duplicates, and low-quality keywords before analysis.

Semantic Audit Pipelines

Patternless Frequency (irregular intervals)

Patternless Frequency is a publishing strategy that uses deliberately irregular intervals for content publication to avoid bot-like patterns.

Semantic Audit Pipelines

Pipeline Persistence

Pipeline Persistence is the practice of saving pipeline step outputs to disk, enabling resuming after failure, debugging, and auditing.

Semantic Audit Pipelines

Pipeline Resumability

Pipeline resumability enables continuing from the point of interruption: each step saves its output to files for recovery.

Semantic Audit Pipelines

Plan-Audit-Improve Cycle

Plan-Audit-Improve Cycle is an iterative semantic audit workflow: plan content → audit existing → improve and fill gaps → plan next.

Semantic Audit Pipelines

Publication Strategy

Publication Strategy is a content publishing plan based on Topical Map sequencing: SEED pages (pillars) first, then NODE pages (supporting content).

Semantic Audit Pipelines

Quality Report

Quality Report is a validation checkpoint after each pipeline step that catches anomalies and decides whether to proceed or fix issues first.

Semantic Audit Pipelines

Semantic Audit Pipeline

Semantic Audit Pipeline is a complete automated process from site crawling to audit report: combining embeddings, clustering, and gap analysis.

Semantic Audit Pipelines

SERP Grounding (CONFIRMED/PREDICTED/SERP-ONLY)

SERP Grounding tags sub-queries by their source: CONFIRMED (LLM + SERP), PREDICTED (LLM only), SERP-ONLY (SERP data only) for reliability assessment.

Semantic Clustering

Attribute Types (Main / Derived / Minor)

Attribute Types (Main / Derived / Minor) classify attributes in a Topical Map by priority: Main (primary), Derived (secondary), Minor (peripheral).

Semantic Clustering

Canonical Query

Canonical Query — the main query representing a thematic cluster, equivalent to a 'primary keyword' but selected based on semantics.

Semantic Clustering

Cascading Clustering (E-commerce)

Cascading Clustering (E-commerce) — Multi-level clustering for e-commerce where categories, subcategories and products form cascading clusters.

Semantic Clustering

Cluster Naming (Central Entity)

Cluster Naming (Central Entity) identifies the Central Entity and Canonical Query for each cluster in the third step of clustering.

Semantic Clustering

Cluster Validation (SERP Overlap)

Cluster Validation (SERP Overlap) validates keyword clusters by measuring SERP overlap between canonical queries and sampled cluster keywords.

Semantic Clustering

Content Gap Detection

Content Gap Detection — The final clustering pipeline stage identifying missing topics by comparing site content against the full topical map.

Semantic Clustering

Content Gap Prioritization (P1–P4)

Content Gap Prioritization (P1–P4) ranks content gaps by their impact: P1 (critical, high volume), P2 (important), P3 (supporting), P4 (nice-to-have).

Semantic Clustering

Correlated Queries (Associations)

Correlated Queries are queries frequently searched by the same users, indicating semantic associations between content clusters on a site.

Semantic Clustering

DBSCAN

DBSCAN is a density-based clustering algorithm that automatically detects cluster count and identifies outliers without specifying k upfront.

Semantic Clustering

Hierarchical Clustering

Hierarchical Clustering creates a tree (dendrogram) of cluster relationships—useful when you don't know the optimal number of groups upfront.

Semantic Clustering

K-means

K-means is a clustering algorithm that divides data points into k groups based on distance from centroids—requires specifying cluster count upfront.

Semantic Clustering

Keyword Clustering (Embeddings + K-means)

Keyword Clustering (Embeddings + K-means) — Second stage in clustering pipeline that groups keywords using embeddings and K-means algorithm.

Semantic Clustering

Query Paths (Search Sequences)

Query Paths are search sequences showing typical user query order — they signal the need for internal linking between clusters.

Semantic Clustering

Query Semantics Google

Query Semantics Google analyzes search query patterns including Query Paths, Correlated Queries, and Sequential Queries for behavioral insights.

Semantic Clustering

Sequential Queries (Query Series)

Sequential Queries are a series of searches performed in order within a single session, revealing user journey and guiding internal linking.

Semantic Clustering

SERP Coherence

SERP Coherence measures a cluster's internal consistency by comparing SERP results of random keywords from the cluster with the canonical query.

Semantic Clustering

SERP Enrichment

SERP Enrichment augments seed keywords with real Google SERP data (PAA, Related Searches, Refine Chips, Filter Sidebar) before LLM expansion.

Semantic Clustering

SERP Hop

SERP Hop is a keyword expansion technique that extracts Related Searches from Google SERPs and uses them as seeds for the next discovery round.

Semantic Clustering

SERP Intelligence

SERP Intelligence analyzes content formats preferred by Google based on SERP data — tells you WHAT KIND of content to create, not just WHAT TOPIC.

Semantic Clustering

SERP Overlap

SERP Overlap measures the percentage of shared results in the top 10 Google results for two queries—overlap above 50% indicates clusters should merge.

Semantic Clustering

Silhouette Score (Selecting k)

Silhouette Score is a clustering quality metric (scale -1 to 1) that automatically determines the optimal number of clusters k.

Semantic Clustering

Token Insertion

Token Insertion is a keyword expansion technique that inserts additional words or phrases into a seed phrase to generate long-tail variants.

Semantic Clustering

Topical Mapping (CORE/OUTER)

Topical Mapping (CORE/OUTER) assigns clusters to CORE (primary) or OUTER (supporting) sections based on attribute classification.

Special Strategies

Context Engineering

Context Engineering is the art of designing in-depth context for AI agents: covering config files, skills, structured data, and memory.

Special Strategies

Domain Classification by Google

Domain Classification by Google assigns domains to topical categories, influencing ranking potential for specific query types and authority assessment.

Special Strategies

Local SEO (law firms)

Local SEO for law firms applies semantic SEO principles to local search optimization — Google My Business, NAP consistency, and client reviews.

Special Strategies

Structure.md (site map for agent)

Structure.md (site map for agent) — Descriptive file with site structure tree created by agent based on sitemap for informed linking.

Special Strategies

YMYL (finance)

YMYL (finance): 'Your Money Your Life' content category covering finance, health, law; Google applies elevated E-E-A-T criteria and verification.

Theoretical Foundations

Attribute (Atrybut)

Attribute is a property describing an entity in EAV: e.g., for entity 'coffee' the attribute is 'brewing method', value 'espresso' or 'drip'.

Theoretical Foundations

BM25 (Saturation and Length)

BM25 (Saturation and Length) is an advanced ranking algorithm accounting for term saturation and length penalties in document scoring.

Theoretical Foundations

Central Entity (CE)

Central Entity (CE) is the main entity (central theme) of a website around which the entire attribute structure is built.

Theoretical Foundations

Central Search Intent (CSI)

Central Search Intent (CSI) is the fundamental goal driving a user's search query—what the user truly wants to accomplish.

Theoretical Foundations

CORE Section (CSI Core)

CORE Section is the main part of a Topical Map, containing topics that represent the primary entity attributes related to Central Search Intent.

Theoretical Foundations

Crawl-Index-Rank (Three-Stage Process)

Crawl-Index-Rank is Google's three-stage process: crawling pages (collection), indexing (cataloging), and ranking by quality.

Theoretical Foundations

Depth (Content Depth)

Depth (Content Depth) measures how thoroughly each aspect of a topic is covered; one of the three pillars of Topical Authority.

Theoretical Foundations

EAV Model

EAV Model is a framework describing any topic as Entity-Attribute-Value triples, forming the foundation of semantic SEO content strategy.

Theoretical Foundations

Entity

Entity is a distinct thematic object (person, thing, concept) that search engines recognize as an independent unit of knowledge with attributes.

Theoretical Foundations

Historical Data (User Signals)

Historical Data refers to user behavior metrics (CTR, time on page, returns) that Google uses to evaluate content quality and determine rankings.

Theoretical Foundations

Hybrid Retrieval

Hybrid Retrieval combines lexical retrieval (BM25, word matching) with semantic retrieval (embeddings, meaning understanding) used by Google.

Theoretical Foundations

Information Retrieval

Information Retrieval is the study of finding relevant information in large document collections, forming the foundation of search engines and AI Search.

Theoretical Foundations

Inverted Index

Inverted Index is a data structure that maps each word to documents containing it — the foundation enabling fast search retrieval.

Theoretical Foundations

Lexical Retrieval

Lexical Retrieval is a search method based on exact word and phrase matching (BM25, TF-IDF) — fast and cheap but can't understand synonyms.

Theoretical Foundations

Momentum (Publishing Pace)

Momentum (Publishing Pace): The pace and regularity of new content publication. Google rewards sites that actively expand topical coverage.

Theoretical Foundations

Multi-Entity

Multi-Entity is a situation where a website processes multiple related entities, each requiring separate EAV analysis and attribute coverage.

Theoretical Foundations

MUM (Multimodal)

MUM (Multimodal) is Google's Multitask Unified Model that processes text, images, and video content while supporting 75 languages.

Theoretical Foundations

Neural Matching (2018)

Neural Matching (2018) uses neural networks to match search queries with relevant pages based on conceptual meaning, not just keyword matching.

Theoretical Foundations

NLP (Natural Language Processing)

NLP (Natural Language Processing) is the AI field that enables machines to understand human language — the foundation of search engines and AI Search.

Theoretical Foundations

OUTER Section (Supporting Topics)

OUTER Section (Supporting Topics) — Part of Topical Map containing supporting topics indirectly related to Central Search Intent.

Theoretical Foundations

Passage Ranking

Passage Ranking is Google's mechanism that identifies and ranks specific page fragments independently from the overall document.

Theoretical Foundations

Popularity

Popularity measures search demand for attributes: High Popularity indicates an attribute warrants dedicated content that responds to real user needs.

Theoretical Foundations

Predicate (CSI Action Verb)

Predicate is the action verb that defines what action users want to perform with an entity — E.g., in 'how to brew coffee' the predicate is 'brew'.

Theoretical Foundations

Prominence

Prominence is a metric that measures how often an attribute appears in top search results for queries related to a Central Entity.

Theoretical Foundations

RankBrain (2015)

RankBrain (2015) is Google's first machine learning system that interprets new queries by matching them to similar past searches.

Theoretical Foundations

RARE (Additional Value)

RARE (Additional Value) — Additional attribute (RARE) in EAV model — gives competitive advantage because most competitors skip it.

Theoretical Foundations

Relevance

Relevance is the primary criterion in attribute filtering that measures how strongly an attribute relates to the Central Search Intent in semantic SEO.

Theoretical Foundations

ROOT (Entity Definition)

ROOT (Entity Definition) — An essential attribute type in the EAV model that defines core characteristics without which an entity loses meaning.

Theoretical Foundations

RPP Attribute Filtering

RPP Attribute Filtering selects entity attributes using three criteria: Relevance, Prominence, and Popularity—only those meeting all qualify.

Theoretical Foundations

Semantic Retrieval

Semantic Retrieval is a search method based on embeddings and meaning understanding — it matches documents based on concepts.

Theoretical Foundations

Semantic SEO

Semantic SEO optimizes websites by assigning meaning to entire site structure rather than targeting individual keywords.

Theoretical Foundations

Source Context (SC)

Source Context (SC) is the perspective from which a website presents its topic: influences classification of attributes as Main.

Theoretical Foundations

TF-IDF (Term Frequency)

TF-IDF (Term Frequency) measures word importance in a document: the more frequent in that document (TF) and rarer in the entire collection (IDF).

Theoretical Foundations

Topical Authority

Topical Authority measures a website's thematic expertise: Google's evaluation of whether a site fullly covers a topic (breadth + depth + pace).

Theoretical Foundations

Topical Coverage

Topical Coverage is how detailedly a website covers a topic: measured by breadth (Vastness), depth (Depth), and publishing pace (Momentum).

Theoretical Foundations

Topical Map

Topical Map: A content organization structure divided into CORE sections (main topics) and OUTER sections (supporting topics).

Theoretical Foundations

UNIQUE (Entity Differentiator)

UNIQUE (Entity Differentiator): An attribute that distinguishes an entity from similar ones, making it unique and recognizable.

Theoretical Foundations

Value

Value (Value) is the specific data assigned to an entity's attribute in the EAV model—e.g., 'brewing temperature' attribute has value '93°C'.

Theoretical Foundations

Vastness (Coverage Breadth)

Vastness (Coverage Breadth): How many different aspects (attributes) of a topic are covered; one of three Topical Coverage pillars.

Theoretical Foundations

Vocabulary Mismatch

Vocabulary Mismatch: problem where users and authors use different words for the same concept — a limitation of lexical retrieval.

Theoretical Foundations

Web Entity

Web Entity is an entity existing in internet space: a topic or brand representation recognizable by Google and AI Search engines.