Module 2: Phrase-Based Indexing

Critical 15 Patents 2 hours

Anna Patterson built the largest search engine of the 21st century. Her phrase-based patents revolutionized how Google understands content and are essential reading for anyone serious about SEO.

Patents Covered

First Generation (Filed July 26, 2004)

#	Patent	Key Innovation
3	US 7,536,408	Core Phrase-Based Indexing
4	US 7,711,679	Duplicate Detection
5	US 7,702,618	Document Version Archiving
6	US 7,603,345	Spam Detection
7	US 9,990,421	Phrase-Based Searching (Updated 2018)
8	US 7,584,175	Document Descriptions
9	US 7,580,929	Personalization
10	US 7,580,921	Phrase Identification
11	US 7,567,959	Multiple Index Retrieval
12	US 7,426,507	Automatic Taxonomy

Second Generation (Filed March 30, 2007)

#	Patent	Key Innovation
13	US 7,693,813	Index Server Architecture
14	US 7,702,614	Index Updating
15	US 7,925,655	Query Scheduling
16	US 20090070312A1	External Phrase Integration

Third Generation

#	Patent	Key Innovation
17	US 8,078,629	Updated Spam Detection

---

Core Concept: Phrase-Based Indexing

Patent: US 7,536,408Inventor: Anna L. Patterson The most important SEO patent you've never read.

What Is a "Good Phrase"?

A phrase is "good" when its component terms co-occur more frequently than random chance would predict.

Example:

• "machine learning" = Good phrase (terms appear together far more than chance)
• "the and" = Not a phrase (appears together often but randomly)

How Phrase-Based Indexing Works

1. Identify Valid Phrases

   • Statistical analysis of term co-occurrence
   • Phrases must exceed randomness threshold

2. Identify Related Phrases

   • Phrases that PREDICT other phrases
   • If a document has "machine learning", it's likely to have "neural networks"

3. Index by Phrase Relationships

   • Documents are scored by their phrase coverage
   • More related phrases = higher topical relevance

The Ranking Formula

Critical Patent Claim (US 9,990,421)

"Documents with MORE RELATED phrases rank higher than those with fewer."

This is the mathematical foundation for topical authority.

Practical Application: Building Phrase Architecture

Step 1: Identify Your Core Phrases

Primary: "content marketing"
Related: "content strategy", "marketing funnel", "audience engagement"
Predictive: "blog posts", "social media", "lead generation"

Step 2: Map Phrase Relationships

Step 3: Ensure Coverage Every page targeting "content marketing" should naturally include phrases from each branch.

---

Patent 6: Spam Detection (US 7,603,345)

How Google Detects Phrase Spam

Spam Indicators

1. High frequency of exact phrases without natural variation
2. Missing related phrases that should naturally appear
3. Unnatural phrase combinations (phrases that don't normally co-occur)
4. Template patterns across multiple pages

Safe Practice

Write for Phrase Completeness, Not Density

Instead of repeating "best coffee maker" 15 times, naturally include:

• "coffee maker reviews"
• "drip coffee machines"
• "brewing quality"
• "carafe capacity"
• "programmable features"

---

Patent 12: Automatic Taxonomy Generation (US 7,426,507)

How Google Builds Topic Hierarchies

SEO Implication

Google understands topic hierarchies. Your site structure should mirror semantic relationships.

Good Structure:

/cooking/
  /cooking/baking/
    /cooking/baking/bread-recipes/
    /cooking/baking/pastry-techniques/
  /cooking/grilling/

Bad Structure:

/page1/
/page2/
/bread-stuff/
/random-grilling/

---

Phrase-Based Personalization (US 7,580,929)

Google personalizes results based on user's phrase history.

What This Means

• Users searching phrase clusters get personalized results
• Historical phrase patterns influence future results
• Building phrase authority helps in personalized contexts

---

Practical Checklist

Content Creation

• [ ] Identify 5-10 related phrases for your target topic
• [ ] Ensure natural coverage of related phrases
• [ ] Include predictive phrases (what else would this content discuss?)
• [ ] Vary exact phrase usage naturally
• [ ] Check for missing expected phrases

Site Architecture

• [ ] Structure reflects topic hierarchies
• [ ] Related content is properly linked
• [ ] Taxonomy matches semantic relationships
• [ ] Phrase clusters are grouped logically

Quality Control

• [ ] No unnatural phrase repetition
• [ ] Related phrases present throughout
• [ ] Natural language flow maintained
• [ ] No template patterns across pages

---

Module Quiz

Question 1: What makes a phrase "good" according to Patterson's patents?

Answer: A phrase is "good" when its component terms co-occur more frequently than random chance would predict.

Question 2: What is the key ranking claim from US 9,990,421?

Answer: "Documents with MORE RELATED phrases rank higher than those with fewer."

Question 3: What three things indicate phrase spam?

Answer:

1. Unnatural co-occurrence patterns
2. Missing related phrases that should appear
3. Template patterns across multiple pages

---

Key Takeaways

1. Phrases, not keywords - Google indexes by meaningful phrases
2. Related phrases matter - Coverage of related phrases determines topical relevance
3. Spam is detectable - Unnatural phrase patterns trigger spam detection
4. Taxonomy is automatic - Google builds topic hierarchies from phrase analysis

---

Next Steps

Continue to Module 3: Entity & Knowledge Graph →

Learn how entities and their relationships power modern search ranking.