Entity Extraction & Content Audit

Source: Bill Slawski, SEO by the Sea — 82 articles on Entity SEO and Knowledge Graph patents Related Patents: Knowledge Graph entity recognition, Named Entity Recognition (NER), entity disambiguation, entity relationship mapping

What Are Entities?

In Google's language, an entity is any "thing" that can be clearly and uniquely identified. Entities are the nouns of Google's understanding model:

• Persons: Bill Slawski, Navneet Panda, Anna Patterson
• Organizations: Google, SEO by the Sea, Moz
• Places: Miami, Florida, the United States
• Concepts: PageRank, E-E-A-T, semantic search
• Products: Google Search Console, Ahrefs, Screaming Frog
• Events: Panda update, Google I/O, the Super Bowl
• Creative works: The Anatomy of a Large-Scale Hypertextual Web Search Engine

Google's systems recognize entities in your content and use them to determine:

1. What your content is about (topic classification)
2. Which Knowledge Graph nodes to associate with your content
3. Whether your content matches the entity-based meaning behind search queries

Entity Types

Named entities: Specific proper nouns (Microsoft, Barack Obama, Paris) Nominal entities: Concept nouns that refer to entities (the president, the company, the algorithm) Pronominal entities: Pronouns that reference previously mentioned entities (he/she/it/they)

Google's NER system tracks all three types and builds entity chains through a document.

The 7-Point Entity Audit

Point 1: Entity Inventory & Classification

Goal: Catalog every entity in your content and classify it by type.

Process:

1. Extract all potential entities from your content (use a tool like TextRazor, Dandelion API, or read manually)
2. Classify each entity: Person / Organization / Place / Concept / Product / Event / Creative Work
3. Note the Knowledge Graph status of each entity — is it in Google's KG? (Search the entity name in Google and check for a Knowledge Panel)
4. Note the Wikipedia status — does the entity have a Wikipedia page? (Wikidata/Wikipedia is a primary KG source)

Output: Entity inventory table with type, KG status, and Wikipedia presence.

Key question for each entity: Is this entity clearly identified by its Wikipedia/KG definition in context, or could it be confused with another entity of the same name?

---

Point 2: First-Mention Contextualization

Goal: Ensure every entity is properly introduced on its first mention.

Undefined entities confuse Google's disambiguation systems. If you mention "Panda" without context, Google has to guess: the animal, the Panda update, the movie, a brand?

The first-mention rule:

• Every entity should be defined or contextualized on first use
• Include attributes that anchor the entity to its correct Knowledge Graph node
• For people: full name + role/credential on first mention
• For organizations: full name + brief descriptor on first mention
• For concepts: define the concept in context, not just drop the term

Good first-mention example: "Bill Slawski, who spent 20 years analyzing Google patents at SEO by the Sea, documented the entity recognition system..."

Weak first-mention example: "Slawski documented the entity recognition system..."

The second version leaves "Slawski" ambiguous — which Slawski? What is SEO by the Sea? The system has to work harder to connect this reference to the correct Knowledge Graph node.

---

Point 3: Entity Attribute Coverage

Goal: Ensure content covers the key attributes Google associates with each important entity.

Every Knowledge Graph entity has a set of attributes (properties) that define it. For a person: name, birthdate, nationality, profession, notable works, organizations associated with. For a company: name, founding date, industry, founders, products, headquarters.

When your content mentions an entity and discusses its key attributes, you're reinforcing the entity-content association. When you mention an entity without developing its key attributes, the association is weak.

Audit method:

1. For each primary entity in your content: search Google and look at its Knowledge Panel
2. Note the top 5-7 attributes listed in the panel
3. Check: does your content address at least 2-3 of these attributes?
4. If content mentions the entity but discusses none of its key attributes — the association is thin

Example: Writing about "Google's Panda update" but only mentioning that it existed. Missing: when it launched (2011), what it targeted (low-quality content), how it worked (site-level quality score), how it changed SEO.

---

Point 4: Entity Disambiguation

Goal: Ensure Google associates your content with the CORRECT entity, not a homonym.

Disambiguation failure is when Google thinks your content is about Entity A when you meant Entity B. This can cause your content to rank for the wrong queries, or fail to rank for the intended ones.

High-risk disambiguation scenarios:

• Brand names that are also common words (Apple, Amazon, Oracle, Slack, Base)
• Person names shared by multiple notable people (Bill Gates — Microsoft founder or NFL coach?)
• Geographic names that appear in multiple places (Springfield, Portland, Dublin)
• Technical terms with consumer meanings (Python, cloud, cookies, crawling)
• Historical vs. current entities (the Roman Mercury vs. Mercury the car brand)

Disambiguation techniques:

1. Type assertion: "Python (the programming language)" or "Python programming language"
2. Attribute anchoring: Include attributes that uniquely belong to the intended entity
3. Co-occurring entity signals: Mention related entities that only co-occur with the intended entity (mentioning "Flask," "Django," "pip" clearly disambiguates Python as the language)
4. Explicit category language: "As a search engine optimization strategy..." makes clear the context is SEO, not finance

---

Point 5: Entity Relationship Mapping

Goal: Ensure your content makes entity relationships explicit, not just lists isolated entities.

Knowledge Graph is built on relationships between entities, not just entities themselves. Relationships like:

• works-for (Bill Slawski works-for Go Fish Digital)
• founded-by (Google founded-by Larry Page and Sergey Brin)
• instance-of (Panda is-an algorithm-update)
• located-in (SEO by the Sea located-in San Diego)

When your content expresses these relationships explicitly, you're helping Google map the content to the correct Knowledge Graph subgraph.

Audit check:

• Does your content just list entities, or does it express how they relate to each other?
• Are relationships explicit ("Google's Panda algorithm, developed by Navneet Panda...") or implicit?
• Do you use relationship language: "developed by," "part of," "related to," "works with," "launched in"?

---

Point 6: Structured Data Alignment

Goal: Ensure schema markup matches the entity claims made in your content.

Schema markup is the machine-readable layer of entity association. When schema markup conflicts with content, it creates a trust deficit.

Alignment checks:

• Author claimed in content matches author in Article schema
• Organization name in content matches name in Organization schema
• Location in content matches address in LocalBusiness schema
• Event dates in content match startDate/endDate in Event schema
• Product features described in content match Product schema properties

Common misalignment errors:

• Schema author name uses formal business name, content uses informal name
• Address abbreviations differ between content and schema
• Schema claims attributes not supported by page content

---

Point 7: Entity Density & Distribution

Goal: Ensure entities are distributed throughout content, not front-loaded or missing from key sections.

Entity signals are stronger when entities appear throughout a document rather than clustered only in one section. This is analogous to the phrase distribution principle from the Phrase-Based Optimizer audit.

Audit method:

• Divide content into thirds (beginning, middle, end)
• Check entity distribution across thirds
• Primary entities should appear in all three thirds
• Supporting entities should appear where contextually relevant — not artificially inserted

Density concerns:

• Too sparse: Entity mentioned once in 2,000 words — association signal is weak
• Too dense: Same entity mentioned 15× in 500 words — over-optimization signal
• Right balance: Primary entity appears every 300-500 words naturally across the document

Entity Optimization Scoring

Dimension	Score (1-10)
Entity inventory completeness	/10
First-mention contextualization	/10
Attribute coverage for primary entities	/10
Disambiguation quality	/10
Relationship mapping clarity	/10
Structured data alignment	/10
Entity density & distribution	/10
TOTAL	/70

Scoring:

• 56-70: Strong entity optimization — content is well-positioned for Knowledge Graph association
• 42-55: Moderate — gaps in attribute coverage or disambiguation need attention
• 28-41: Weak — entity signals are unclear, disambiguation failures likely
• Below 28: Entity optimization not present — Google may misclassify the content entirely

Tools for Entity Analysis

• TextRazor: Entity extraction API with Knowledge Graph linking
• Dandelion API: Named entity recognition with Wikipedia entity linking
• Google's Natural Language API: Salience scores for entities in content
• InLinks.net: Entity-focused SEO optimization platform
• Google Rich Results Test: Verify schema entity markup is valid