version: "1.0.1"

SKILL: research/entity-extraction.skill

SYSTEM: Nexus — Modular AI-Powered SEO Operating System

VERSION: 1.0.0

LAYER: Research / Semantic Intelligence

EXECUTION PRIORITY: PARALLEL OR POST-SERP — Runs alongside or after deep-serp-analysis.skill.

POSITION: Semantic backbone builder between SERP intelligence and content generation.

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

SECTION 1 — ROLE

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

This skill is the semantic entity intelligence engine of the Nexus system.

It extracts, maps, classifies, scores, and prioritizes the named entities and conceptual units that define a topic's semantic landscape — the entities that Google's Knowledge Graph, topical authority algorithms, and answer engines use to determine whether a piece of content genuinely understands its subject matter at depth.

Modern search ranking is not merely about keyword presence. Google evaluates content through entity co-occurrence patterns: does this content mention the tools, experts, organizations, processes, metrics, and concepts that belong to the topic's semantic field? When these entities are present, correctly contextualized, and related to each other appropriately, the content signals topical authority. When they are absent or incorrectly framed, the content signals surface-level treatment — regardless of word count or structure.

This skill builds the semantic backbone that downstream content skills use to ensure every piece of content contains the full entity coverage required to compete as a topically authoritative resource.

Primary Responsibilities

What This Skill Is NOT

• It is not a keyword extractor. Keywords and entities are distinct. A keyword is what users search for. An entity is a named thing — a concept, tool, person, organization, or process — that the topic requires understanding.
• It is not a content generator. It produces entity intelligence — content-generation-engine.skill uses that intelligence to write.
• It is not a SERP analyzer. It receives SERP data from upstream skills — it does not perform page analysis itself.
• It is not a link builder. Entity relationship mapping is semantic, not structural — it does not describe hyperlink relationships.
• It is not optional for technical or expert-level content. For any content targeting audiences with domain expertise, entity coverage is as important as keyword targeting.

The Foundational Principle of This Skill

An entity is not relevant because it appears in the content.

An entity is relevant because it belongs to the semantic field of the topic — because a genuine expert writing about this topic would naturally include it.

This skill's job is to identify all such entities from the SERP evidence and user content gaps,

then classify each by how essential it is to topical authority.

Every entity in the output must be there because it serves the reader's understanding

or the content's semantic completeness — not because it appeared once in a search result.

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

SECTION 2 — INPUT

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Primary Inputs

Optional Inputs

Input Confidence Levels

The quality and completeness of entity extraction directly depends on input richness. Track and report input confidence:

Input Pre-Conditions

1. Is `deep-serp-analysis.skill` data available? If yes → use as primary entity source. If no → operate in inference mode; flag all output.
2. Is `keyword-discovery.skill` output available? If yes → load full keyword universe for semantic neighbor expansion. If no → derive entity candidates from topic alone.
3. Is user existing content available for gap analysis? If yes → run full gap detection in Step 5. If no → run SERP-only entity map with note that gap analysis was skipped.
4. Has entity extraction been run before for this keyword/topic? → Query memory-controller.skill. If within freshness threshold (14 days) → offer cached. If stale → proceed fresh.

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

SECTION 3 — OUTPUT

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Primary Output — Structured Entity Intelligence Report

NEXUS ENTITY INTELLIGENCE REPORT
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Session ID              : [ID]
Target Topic / Keyword  : [keyword or topic]
Input Confidence        : [HIGH / MEDIUM-HIGH / MEDIUM / LOW]
Total Entities Extracted: [count]
  Critical              : [count]
  Important             : [count]
  Optional              : [count]
SERP Pages Referenced   : [count]
Gap Analysis Run        : [YES / NO — reason if no]
Inference Flags         : [count of inferred entities]
Generated By            : research/entity-extraction.skill
Timestamp               : [ISO 8601]
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
 
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
PART 1 — FULL ENTITY TABLE
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
 
Grouped by Entity Type:
 
─────────────────────────────────────────────────────────────────────
CATEGORY 1: TOOLS / SOFTWARE
─────────────────────────────────────────────────────────────────────
 
| # | Entity             | Type     | Relevance    | Frequency | Priority  | AEO Tag | Why It Matters                                          |
|---|--------------------|----------|--------------|-----------|-----------|---------|----------------------------------------------------------|
| 1 | [Entity Name]      | Tool     | High         | [N/N pgs] | Critical  | DEF     | [specific reason tied to topic and audience]            |
| 2 | [Entity Name]      | Software | Medium       | [N/N pgs] | Important | FAQ     | [specific reason]                                       |
| 3 | [Entity Name]      | Tool     | Low          | [N/N pgs] | Optional  | —       | [specific reason]                                       |
...
 
─────────────────────────────────────────────────────────────────────
CATEGORY 2: TECHNOLOGIES
─────────────────────────────────────────────────────────────────────
 
| # | Entity             | Type       | Relevance | Frequency | Priority  | AEO Tag | Why It Matters                                         |
|---|--------------------|------------|-----------|-----------|-----------|---------|--------------------------------------------------------|
| 1 | [Entity Name]      | Technology | High      | [N/N pgs] | Critical  | DEF     | [reason]                                               |
...
 
─────────────────────────────────────────────────────────────────────
CATEGORY 3: CONCEPTS / METHODS
─────────────────────────────────────────────────────────────────────
 
| # | Entity             | Type    | Relevance | Frequency | Priority  | AEO Tag | Why It Matters                                         |
|---|--------------------|---------|-----------|-----------|-----------|---------|--------------------------------------------------------|
| 1 | [Entity Name]      | Concept | High      | [N/N pgs] | Critical  | DEF+FAQ | [reason]                                               |
...
 
─────────────────────────────────────────────────────────────────────
CATEGORY 4: PEOPLE / EXPERTS
─────────────────────────────────────────────────────────────────────
 
| # | Entity             | Type   | Relevance | Frequency | Priority  | AEO Tag | Why It Matters                                         |
|---|--------------------|--------|-----------|-----------|-----------|---------|--------------------------------------------------------|
| 1 | [Entity Name]      | Expert | High      | [N/N pgs] | Important | QUOTE   | [reason]                                               |
...
 
─────────────────────────────────────────────────────────────────────
CATEGORY 5: ORGANIZATIONS
─────────────────────────────────────────────────────────────────────
 
| # | Entity             | Type         | Relevance | Frequency | Priority  | AEO Tag | Why It Matters                                     |
|---|--------------------|--------------|-----------|-----------|-----------|---------|-----------------------------------------------------|
| 1 | [Entity Name]      | Organization | High      | [N/N pgs] | Critical  | REF     | [reason]                                            |
...
 
─────────────────────────────────────────────────────────────────────
CATEGORY 6: PROCESSES / WORKFLOWS
─────────────────────────────────────────────────────────────────────
 
| # | Entity             | Type     | Relevance | Frequency | Priority  | AEO Tag | Why It Matters                                     |
|---|--------------------|----------|-----------|-----------|-----------|---------|-----------------------------------------------------|
| 1 | [Entity Name]      | Process  | High      | [N/N pgs] | Critical  | HOW     | [reason]                                            |
...
 
─────────────────────────────────────────────────────────────────────
CATEGORY 7: METRICS / KPIs
─────────────────────────────────────────────────────────────────────
 
| # | Entity             | Type   | Relevance | Frequency | Priority  | AEO Tag | Why It Matters                                     |
|---|--------------------|--------|-----------|-----------|-----------|---------|-----------------------------------------------------|
| 1 | [Entity Name]      | Metric | High      | [N/N pgs] | Critical  | DEF     | [reason]                                            |
...
 
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
PART 2 — ENTITY GAP ANALYSIS
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
 
[Only produced if user existing content is available]
 
MISSING ENTITIES (present in SERP / absent in user content):
| Entity             | Type     | Priority  | SERP Frequency | Gap Severity   |
|--------------------|----------|-----------|----------------|----------------|
| [Entity Name]      | [type]   | Critical  | [N/N pgs]      | CRITICAL GAP   |
| [Entity Name]      | [type]   | Important | [N/N pgs]      | SIGNIFICANT    |
| [Entity Name]      | [type]   | Optional  | [N/N pgs]      | MINOR          |
 
UNDERUSED ENTITIES (present in user content but insufficiently developed):
| Entity             | Type     | Priority  | User Content Depth | Required Depth |
|--------------------|----------|-----------|--------------------|----------------|
| [Entity Name]      | [type]   | Important | MENTIONED          | DEFINED        |
| [Entity Name]      | [type]   | Critical  | DEFINED            | EXPLAINED+EXAMPLE |
 
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
PART 3 — RELATIONSHIP MAP
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
 
[Entity relationship chains — structured by relationship type]
 
DEPENDENCY RELATIONSHIPS (A depends on / requires B):
  → [Entity A] — requires → [Entity B]
    Reason: [why A cannot function without B]
  → [Entity A] — requires → [Entity B]
  ...
 
FUNCTIONAL RELATIONSHIPS (Tool used in Process):
  → [Entity: Tool] — used in → [Entity: Process]
    Context: [how the tool is used in this process]
  → [Entity: Tool] — used in → [Entity: Process]
  ...
 
MEASUREMENT RELATIONSHIPS (Metric measures Outcome):
  → [Entity: Metric] — measures → [Entity: Outcome/Concept]
    Context: [what this metric quantifies and why it matters]
  ...
 
HIERARCHICAL RELATIONSHIPS (A is a type of / subcategory of B):
  → [Entity: Specific] — is a type of → [Entity: General]
  ...
 
ENABLING RELATIONSHIPS (A enables / produces B):
  → [Entity: Process/Method] — produces → [Entity: Outcome/Metric]
  ...
 
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
PART 4 — AEO ENTITY MAP
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
 
DEFINITION TARGETS (DEF — suitable for featured snippet paragraph):
  | Entity           | Recommended Definition Format | Target Query                          |
  |------------------|-------------------------------|---------------------------------------|
  | [Entity Name]    | [Paragraph / 2-sentence]      | "What is [entity]?"                   |
  | [Entity Name]    | [Paragraph / 2-sentence]      | "What does [entity] mean?"            |
 
FAQ TARGETS (FAQ — suitable for FAQ schema and PAA alignment):
  | Entity           | Recommended Question                     | Answer Format     |
  |------------------|------------------------------------------|-------------------|
  | [Entity Name]    | "[question phrasing]"                    | [list / paragraph]|
  | [Entity Name]    | "[question phrasing]"                    | [format]          |
 
HOW-TO TARGETS (HOW — suitable for HowTo schema and process snippet):
  | Entity           | Recommended Question                     | Steps Required    |
  |------------------|------------------------------------------|-------------------|
  | [Entity Name]    | "How to use/implement [entity]?"         | [N] steps         |
 
REFERENCE TARGETS (REF — suitable for authority signal, cited as source):
  | Entity           | Reference Context                        | Citation Type     |
  |------------------|------------------------------------------|-------------------|
  | [Entity Name]    | [where this entity should be cited]      | [stat / report / standard] |
 
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
PART 5 — ENTITY CLUSTER MAP
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
 
[Grouping entities by semantic proximity across categories]
 
Entity Cluster 1 — [Cluster Label]:
  Core Entity   : [entity]
  Related       : [entity], [entity], [entity]
  Cluster Purpose: [what semantic territory this cluster covers]
  Content Section: [which blueprint section should cover this cluster]
 
Entity Cluster 2 — [Cluster Label]:
  Core Entity   : [entity]
  Related       : [entity], [entity]
  Cluster Purpose: [purpose]
  Content Section: [section reference]
 
...
 
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
PART 6 — PRIORITY SUMMARY TABLE
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
 
CRITICAL ENTITIES (must include — failure to cover signals shallow content):
  [Entity 1] ([Type]) | [Entity 2] ([Type]) | [Entity 3] ([Type]) | ...
 
IMPORTANT ENTITIES (should include — strengthens topical authority):
  [Entity 1] ([Type]) | [Entity 2] ([Type]) | ...
 
OPTIONAL ENTITIES (enhancement — adds depth for expert audiences):
  [Entity 1] ([Type]) | [Entity 2] ([Type]) | ...
 
INFERRED ENTITIES (marked as inferred — confirm before including):
  [Entity 1] ([Type]) — INFERRED: [reason for inference]
  [Entity 2] ([Type]) — INFERRED: [reason]
 
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
DISPATCH READY
Feeds Into              : semantic-clustering.skill
                          content-generation-engine.skill
                          optimization-engine.skill
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Output Field Definitions

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

SECTION 4 — STEP-BY-STEP LOGIC

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

▸ STEP 1 — DEFINE ENTITY CATEGORIES

Before any extraction begins, establish the full taxonomy of entity types this skill will search for across all data sources. This taxonomy ensures no entity category is missed during extraction.

The 7 Entity Categories:

Category 1 — Tools / Software

Definition: Named products, platforms, applications, or digital tools that are directly relevant to the topic.

Detection Signals: Named with capital letters; appear in comparison sections; have version numbers; have official brand names.

Category 2 — Technologies

Definition: Underlying technical standards, architectures, protocols, or technology types that form the infrastructure of the topic's domain.

Detection Signals: Abbreviated names (all-caps or camelCase); appear in technical sections; often co-occur with "based on" or "built on."

Category 3 — Concepts / Methods

Definition: Named theories, methodologies, frameworks, approaches, or conceptual constructs that are central to understanding the topic.

Detection Signals: Appear in educational/definitional sections; frequently appear as H2 or H3 headings; often followed by "is a" or "involves."

Category 4 — People / Experts

Definition: Named individuals who are recognized authorities on the topic — researchers, practitioners, thought leaders, or experts whose work or opinions are relevant.

Detection Signals: Full names with titles; appear in quote sections; cited by name with professional context.

Category 5 — Organizations

Definition: Named companies, institutions, standards bodies, associations, regulatory bodies, or research organizations with a significant relationship to the topic.

Detection Signals: Preceded by articles like "the" or "by"; appear in authority or reference sections; co-occur with "published by," "according to," "certified by."

Category 6 — Processes / Workflows

Definition: Named sequences of steps, operations, or procedures that describe how something is done within the topic's domain.

Detection Signals: Often contain verbs or action words in their names; appear in "how to" and step-by-step sections; frequently described with phase or stage terminology.

Category 7 — Metrics / KPIs

Definition: Named quantitative measurements, performance indicators, or scoring systems used to evaluate outcomes within the topic's domain.

Detection Signals: Appear in measurement or performance sections; co-occur with numbers and percentages; often abbreviated with acronyms.

Category Inventory Initialization:

Before extraction begins, build a Category Inventory — an empty structured container for each of the 7 categories. All entities extracted in Steps 2 and 3 are deposited into the appropriate category container before classification occurs.

▸ STEP 2 — SERP ENTITY EXTRACTION

Extract all candidate entities from the SERP data provided by deep-serp-analysis.skill. Extraction focuses on three content zones where entities carry the strongest topical authority signal.

Phase 2A — Extraction Zones and Their Entity Signals

Zone 1 — Page Titles and H1 Headings:

Entities appearing in page titles and H1 headings are the highest-confidence entities — they were important enough to be used as the defining label for the entire page.

Extraction method:

• Collect all page titles from the analyzed SERP pages.
• Identify named entities (capitalized nouns, named concepts, product names, technical terms).
• Record entity name + page count (how many titles contain this entity).
• Entities in ≥3 titles → classify as HIGH relevance candidates.

Zone 2 — H2 and H3 Headings:

Entities in subheadings reveal the topic's structural semantic field — what sub-topics and named concepts are treated as important enough to warrant their own sections.

Extraction method:

• Collect all H2 and H3 headings from all analyzed pages (from deep-serp-analysis.skill Dimension 1).
• Identify all named entities across the heading set.
• Record entity name + heading count (how many headings contain this entity across all pages).
• Note position: H2 entities carry more weight than H3 entities.

Zone 3 — Intro Sections (First 200 Words):

Entities in introductory sections are the ones authors judge essential to establish from the start — they are the definitional anchors of the topic.

Extraction method:

• From snippet data and intro analysis (where available from deep-serp-analysis.skill).
• Extract entities that appear in the opening content of ranking pages.
• Record entity name + appearance count.

Zone 4 — PAA Questions (from serp-intelligence.skill if available):

PAA questions reveal which entities Google considers relevant enough to surface as follow-up queries. Every named entity in a PAA question is a semantically significant entity for the topic.

Extraction method:

• Parse all PAA questions for named entities.
• Every entity in a PAA question → classify as at minimum MEDIUM relevance.

Phase 2B — Keyword Universe Entity Extraction

Beyond SERP page analysis, extract entities from the keyword universe produced by keyword-discovery.skill.

Extraction method:

• Scan all 8 keyword dimensions for named entities.
• Dimension 3 (Comparison): Named tools and products in "A vs B" keywords are entities.
• Dimension 4 (Alternatives): Named tools in "alternatives to X" keywords are entities.
• Dimension 7 (Semantic Expansion): Tools, processes, and concepts in semantic expansion keywords are entities.
• Dimension 8 (Niche Modifiers): Industry and use-case names become Organization or Process entities.

Phase 2C — Competitor Content Entity Extraction (Optional)

If competitor content is available as a raw text input:

• Parse headings, bold text, and first paragraphs for named entities.
• Entities appearing in multiple competitor pieces are higher-relevance candidates.
• Flag all competitor-derived entities with source: COMPETITOR CONTENT.

Phase 2D — Entity Candidate Pool Assembly

After all three zones are processed:

1. Combine all extracted entity candidates into a single pool.
2. Remove clearly non-entity items (generic adjectives, stop words, numerical values without named context).
3. Group all instances of the same entity (accounting for synonyms and abbreviations — e.g., "Customer Relationship Management software" and "CRM software" → same entity: "CRM").
4. Count total mentions across all zones for each unique entity.
5. Deposit each entity candidate into its Category Inventory container.

▸ STEP 3 — FREQUENCY SCORING (TF-IDF SIMULATION)

Apply a TF-IDF-simulated scoring model to every entity candidate. This model assigns each entity a Relevance level based on how frequently and prominently it appears across the analyzed SERP pages — prioritizing entities that are consistently important over entities that appear sporadically.

Frequency Scoring Framework:

Term Frequency Component (TF):

Count how many of the analyzed SERP pages mention each entity, and in what zone:

Weighted TF Calculation per Entity:

TF_score = Σ(zone_weight × zone_mention_count) across all pages

Inverse Document Frequency Component (IDF — Simulated):

The IDF component penalizes entities that are too generic — entities that would appear in nearly any content on any topic (e.g., "content," "users," "data"). These are not semantically distinctive.

Apply a penalization factor:

Combined Relevance Score:

Relevance_score = TF_score × IDF_multiplier

Relevance Level Classification:

Note: The score boundaries are relative to the extracted entity set — not absolute values. The top third of this specific entity pool is HIGH, regardless of raw score.

Frequency Documentation:

For every entity, record:

• Raw mention count (total across all pages and zones).
• Page count (how many unique pages mention the entity — reported as N/N pages).
• Weighted TF score.
• IDF multiplier applied.
• Final Relevance Level.

▸ STEP 4 — ENTITY CLASSIFICATION

After frequency scoring, formally classify every entity in the candidate pool.

Phase 4A — Entity Type Assignment

Assign every entity to its most precise sub-type within its category:

Phase 4B — Canonical Name Normalization

For every classified entity, establish its canonical name — the single, authoritative form of the entity's name to be used consistently throughout all output:

Rules:

1. Use the entity's official or most widely recognized name.
2. For tools and software → use the official product name (exact capitalization).
3. For acronyms → include the full form alongside the abbreviation on first use: "Customer Acquisition Cost (CAC)".
4. For concepts → use the most established academic or industry term.
5. For people → use full name with professional identifier.
6. Never use informal nicknames as the canonical name.

Phase 4C — Relevance Validation

After classification, validate that every HIGH relevance entity genuinely belongs to the topic's semantic field:

Validation question: "Would a recognized expert writing about [topic] naturally include this entity in their content without being prompted?"

• YES → confirmed as HIGH relevance.
• UNCERTAIN → downgrade to MEDIUM.
• NO → downgrade to LOW or remove entirely.

This validation prevents frequency bias — an entity that appears in many pages due to coincidental co-occurrence rather than genuine semantic relevance.

▸ STEP 5 — ENTITY GAP DETECTION

Compare the SERP entity set against the user's existing content to identify which entities are missing from the user's coverage and which are present but underdeveloped.

Phase 5A — User Content Entity Extraction

If user existing content is available:

1. Extract entities from the user's published articles (titles, headings, and body content).
2. Apply the same 7-category taxonomy.
3. Classify user-present entities by depth of treatment:

Phase 5B — Gap Identification

For each entity in the SERP entity set:

1. Is it present in the user's content? → YES / NO.
2. If YES → at what depth level?
3. If NO → record as MISSING entity.
4. If YES but depth < required level (determined by entity priority) → record as UNDERUSED entity.

Required Depth by Priority:

Phase 5C — Gap Severity Classification

Phase 5D — Gap Prioritization

Rank all gaps by:

1. Entity priority (CRITICAL > IMPORTANT > OPTIONAL).
2. SERP frequency (higher-frequency entities = higher-priority gaps).
3. Gap severity (CRITICAL GAP > UNDERUSE GAP).

The highest-ranking gaps feed directly into the content-generation-engine.skill as the first entities to address.

▸ STEP 6 — RELATIONSHIP MAPPING

Define the semantic relationships between entities. This relationship map is the connective tissue of the topic's semantic architecture — it tells content-generation-engine.skill not just which entities to include, but how they relate to each other in the content.

Phase 6A — Relationship Type Taxonomy

Phase 6B — Relationship Identification Protocol

For every pair of entities where a meaningful relationship exists:

1. Identify the relationship type from the taxonomy.
2. Express it as: [Entity A] — [relationship] → [Entity B].
3. Add a one-sentence context note explaining why this relationship matters for content.
4. Record the relationship in the Relationship Map.

Relationship Identification Guidance by Category:

Phase 6C — Relationship Cluster Formation

After all individual relationships are mapped, identify relationship clusters — groups of entities that are densely interconnected:

1. A relationship cluster exists when 3+ entities are all linked to each other through at least one relationship each.
2. Name each cluster with a descriptive label.
3. Note which content section in the blueprint should cover each relationship cluster.

Relationship Density Rule:

No relationship should be manufactured. Every relationship documented must reflect a genuine semantic or functional connection. If two entities appear in the same topic domain but have no meaningful relationship to each other → do not create a relationship entry.

▸ STEP 7 — AEO ENTITY TAGGING

Identify which entities are suitable targets for answer engine optimization — the structural content elements that Google's featured snippets, AI Overviews, and PAA boxes extract.

Phase 7A — AEO Tag Definitions

Phase 7B — AEO Tagging Protocol

For every entity in the full entity table, evaluate it against each AEO tag:

DEF qualification:

• Is this a named concept, technology, or method that has a clear, searchable definition?
• Does the SERP for "what is [entity]" likely produce a featured snippet?
• Would defining this entity add genuine value to readers at this topic's funnel stage?
• If YES to all three → assign DEF tag.

FAQ qualification:

• Does this entity appear as the subject of a PAA question in serp-intelligence.skill data?
• Is this entity likely to generate question-based searches ("how to use [entity]", "when to use [entity]", "is [entity] worth it")?
• If YES to either → assign FAQ tag.

HOW qualification:

• Is this entity a process, workflow, or method?
• Do users commonly search for implementation guidance for this entity?
• If YES to both → assign HOW tag.

QUOTE qualification:

• Is this entity a named person or organization whose published statements are widely cited in the topic domain?
• Would quoting this entity improve the content's E-E-A-T signals?
• If YES to both → assign QUOTE tag.

REF qualification:

• Is this entity an organization, institution, or study that is recognized as a primary data or authority source for the topic?
• If YES → assign REF tag.

Multi-Tag Assignment: An entity may receive multiple AEO tags if it qualifies for more than one (e.g., a concept that is both worth defining AND worth asking a FAQ question about → DEF + FAQ).

Phase 7C — AEO Entity Integration Guidance

For every AEO-tagged entity, produce a brief integration note:

Entity: [Name]
AEO Tag: [DEF / FAQ / HOW / QUOTE / REF]
Target Query: "[specific search query this entity targets in AEO context]"
Content Placement: [which section of the blueprint should handle this entity's AEO content]
Format Recommendation: [paragraph / numbered list / attribution / table]

▸ STEP 8 — PRIORITIZATION

Assign every entity a final priority classification that drives content inclusion decisions in content-generation-engine.skill.

Phase 8A — Priority Classification Criteria

Phase 8B — Priority Assignment Rules

Priority is determined by the combination of Relevance Level (from Step 3) and Entity Category weight:

*Core categories for most topics: Concepts/Methods, Processes/Workflows, Metrics/KPIs **Secondary categories: People/Experts, Organizations, Tools/Software, Technologies (may be core for technical topics)

Note: Category weight is topic-dependent. For a technical programming topic, Technologies is a core category. For a marketing strategy topic, Concepts/Methods and Metrics/KPIs are core. Adjust category weighting based on topic domain before applying the priority matrix.

Phase 8C — CRITICAL Entity Validation

Before finalizing CRITICAL designations, run a validation check:

For each CRITICAL entity, ask:

1. Does omitting this entity produce a content piece that is clearly incomplete compared to the top-ranking pages? → YES → confirm CRITICAL.
2. Would a domain expert reading the content be surprised that this entity is not mentioned? → YES → confirm CRITICAL.
3. Does the entity appear in the titles or H2 headings of ≥50% of analyzed SERP pages? → YES → confirm CRITICAL.

If an entity does not satisfy at least 2 of the 3 validation checks → downgrade to IMPORTANT.

Phase 8D — Priority Summary Construction

After all entities are assigned priorities:

1. List all CRITICAL entities together (Part 6 of output).
2. List all IMPORTANT entities together.
3. List all OPTIONAL entities together.
4. List all INFERRED entities separately with their inference basis.
5. Calculate counts for the output header.

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

SECTION 5 — ADVANCED LOGIC

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

▸ SEMANTIC RELEVANCE SIMULATION

Beyond frequency scoring, this skill simulates semantic relevance by evaluating each entity against the conceptual field of the topic — not just how often it appears, but how central it is to the topic's meaning.

Semantic Centrality Test:

For each entity, evaluate:

Entities with HIGH semantic centrality AND HIGH topical specificity → elevated to CRITICAL regardless of frequency score.

Entities with LOW topical specificity (generic entities that could belong to any topic) → downgraded by one priority level regardless of frequency.

▸ ENTITY CLUSTER MAPPING

Group entities into semantic clusters — not based on their category type, but based on how they are conceptually related in the content.

Cluster Formation Rules:

1. A cluster forms when ≥3 entities share a primary relationship (e.g., all relate to the same process, or all are tools used in the same workflow).
2. Every entity in the full entity table must be assigned to at least one cluster.
3. An entity may belong to two clusters if it connects two semantic areas of the topic.
4. Clusters should align with content sections in the blueprint — each cluster maps to approximately one H2 section.

Cluster Naming Convention:

Cluster names describe the conceptual territory, not the entity list:

• ✅ "CRM Data Management" (describes the territory)
• ✅ "Sales Performance Measurement" (describes the territory)
• ❌ "Group 1: Metrics and Tools" (generic label)
• ❌ "Cluster A" (no descriptive value)

▸ CONTEXTUAL IMPORTANCE OVERRIDE

Frequency is the primary relevance signal, but contextual importance can override it in specific circumstances.

Override Conditions:

▸ LOW-VALUE ENTITY FILTERING

Prevent the entity table from being diluted by generic, non-informative, or irrelevant entities.

Low-Value Entity Identification:

An entity is low-value if it meets ANY of the following conditions:

• It is a generic noun that could appear in content about any topic (e.g., "users," "teams," "data," "content").
• It is a placeholder name with no recognized canonical form in the domain.
• Its IDF multiplier is 0.2 (universal penalty applied in Step 3).
• It has no meaningful relationship to any other entity in the map (isolated entity with no connections).
• Its "Why It Matters" reason cannot be written in a specific, topic-grounded way.

Low-value entities are:

• Removed from the entity table entirely.
• Logged in the extraction deduplication log.
• Not counted in the output total.

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

SECTION 6 — DEDUPLICATION RULES

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Rule 1 — Canonical Name Deduplication

Every entity must appear exactly once in the output under its canonical name. If the same entity is extracted multiple times under different names (full name, abbreviation, alias, informal name):

1. Identify all name variants.
2. Select the canonical name (most formal, most widely recognized).
3. Merge all variant mentions into the single canonical entity record.
4. Sum all variant mention counts into the merged total.
5. Log all aliases: "CRM = Customer Relationship Management software = Customer Relationship Manager."

Rule 2 — Semantic Entity Deduplication

Two entities that describe the same real-world thing but from different angles are semantic duplicates:

• "Sales Pipeline" and "Sales Funnel" — not identical, but semantically overlapping.
• Detection: if two entities have the same core meaning, the same relationship partners, and the same AEO treatment → merge into the more established term; note the alternative as an alias.

Semantic duplicate test: "Would including both entities in the same sentence feel redundant?" → YES → merge.

Rule 3 — Category Assignment Deduplication

Each entity belongs to exactly one category. If an entity could be classified in two categories (e.g., a named framework that is both a Concept AND a Tool):

• Assign to the category that best represents its primary function in the topic domain.
• Note the secondary classification in the entity's record.
• Do not list the entity in two category sections.

Rule 4 — Relationship Deduplication

Each entity pair has at most one primary relationship in the map. If multiple relationship types could describe the same pair:

• Select the most specific relationship type.
• Note secondary relationship as a parenthetical if meaningful.
• Do not create two separate relationship entries for the same pair.

Rule 5 — Priority Deduplication

Each entity receives exactly one priority classification. Priority cannot be both CRITICAL and IMPORTANT. If scoring produces a borderline result (on the boundary between two levels) → assign the higher priority and document the borderline note.

Rule 6 — Memory Deduplication

Query memory-controller.skill before producing the entity table. If an entity extraction report for this keyword exists within freshness threshold (14 days):

• Load the cached report.
• Present to user with timestamp.
• Ask: "Use existing entity map or refresh?"
• Never store two entity maps for the same keyword in the same freshness window.

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

SECTION 7 — ANTI-FLUFF RULES

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Prohibited Entity Reporting Behaviors

Required Entity Reporting Behaviors

• Every entity record must have all fields populated: Entity, Type, Relevance, Frequency (with zone), Priority, AEO Tag, Why It Matters.
• Every CRITICAL entity must be validated against Phase 8C criteria.
• Every relationship in the map must include a context note.
• Every inferred entity must be explicitly flagged as INFERRED with its inference basis.
• The entity table must be organized by category — not by priority or alphabetical order.

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

SECTION 8 — INTEGRATION WITH OTHER SKILLS

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

▸ deep-serp-analysis.skill

▸ keyword-discovery.skill

▸ serp-intelligence.skill

▸ semantic-clustering.skill

▸ content-generation-engine.skill

▸ optimization-engine.skill

▸ memory-controller.skill

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

APPENDIX A — FAILSAFE LOGIC

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

APPENDIX B — ENTITY CATEGORY QUICK REFERENCE

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

7-CATEGORY ENTITY TAXONOMY
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
CAT 1 — TOOLS / SOFTWARE
  Sub-types: Tool | Software | Plugin | API | Framework(software)
  Signals: Named products; brand names; version numbers
 
CAT 2 — TECHNOLOGIES
  Sub-types: Protocol | Architecture | Language | Standard | Infrastructure
  Signals: All-caps abbreviations; "based on" co-occurrence; technical specifications
 
CAT 3 — CONCEPTS / METHODS
  Sub-types: Concept | Method | Framework(conceptual) | Model | Strategy
  Signals: Definitional sections; "what is"; appears as H2 topic headings
 
CAT 4 — PEOPLE / EXPERTS
  Sub-types: Researcher | Practitioner | Thought Leader | Author
  Signals: Full names; co-occurrence with quotes or attribution; "according to"
 
CAT 5 — ORGANIZATIONS
  Sub-types: Company | Institution | Standards Body | Association | Regulatory Body
  Signals: "The"; "by"; "according to"; "published by"; "certified by"
 
CAT 6 — PROCESSES / WORKFLOWS
  Sub-types: Process | Workflow | Procedure | Pipeline | Lifecycle
  Signals: Action-verb names; "stage"; "phase"; "step"; process-flow headings
 
CAT 7 — METRICS / KPIS
  Sub-types: Metric | KPI | Score | Rate | Index
  Signals: Paired with numbers/percentages; measurement sections; performance headings
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

APPENDIX C — AEO TAG DECISION TREE

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

AEO TAGGING DECISION TREE
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
For each entity, answer in order:
 
1. Is this entity a named concept, technology, or method?
   AND is "what is [entity]" a plausible search query?
   → YES: Add DEF tag. Continue to next question.
   → NO: Skip DEF.
 
2. Does this entity appear in a PAA question?
   OR is "how/when/why [entity]" a plausible search query?
   → YES: Add FAQ tag. Continue.
   → NO: Skip FAQ.
 
3. Is this entity a named process, workflow, or method?
   AND do users search for HOW to implement it?
   → YES: Add HOW tag. Continue.
   → NO: Skip HOW.
 
4. Is this entity a named person or organization?
   AND are their statements widely cited for authority?
   → YES: Add QUOTE tag. Continue.
   → NO: Skip QUOTE.
 
5. Is this entity an organization, institution, or study?
   AND is it commonly cited as a primary source for data or standards?
   → YES: Add REF tag.
   → NO: Skip REF.
 
If no tags were assigned → assign "—" (no AEO targeting).
Multiple tags are permitted if multiple questions answer YES.
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

APPENDIX D — RELATIONSHIP MAPPING QUICK REFERENCE

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

RELATIONSHIP TYPE REFERENCE
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
DEPENDENCY:   [A] — requires → [B]
  Use when: A cannot be implemented or understood without B
 
FUNCTIONAL:   [A] — used in → [B]
  Use when: A (tool/method) is applied within B (process/workflow)
 
MEASUREMENT:  [A] — measures → [B]
  Use when: A (metric) quantifies the performance of B (outcome)
 
HIERARCHICAL: [A] — is a type of → [B]
  Use when: A is a specific instance or subcategory of the broader B
 
ENABLING:     [A] — produces → [B]
  Use when: Performing A (process) generates or creates B (output)
 
CONTRASTING:  [A] — differs from → [B]
  Use when: A and B serve similar purposes but in meaningfully distinct ways
 
SEQUENTIAL:   [A] — precedes → [B]
  Use when: A always occurs before B in a defined sequence or workflow
 
COMPOSITIONAL:[A] — is part of → [B]
  Use when: A is a component, feature, or element of the larger B
 
RULE: One relationship per entity pair.
RULE: Never manufacture relationships — only document genuine connections.
RULE: Every relationship requires a one-sentence context note.
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

SKILL FILE END: research/entity-extraction.skill Nexus SEO Operating System — Version 1.0.0