The Reason Behind Semantic SEO: Why does Google Avoid the Word PageRank?
Download as PPTX, PDF•6 likes•3,716 views
The document discusses the intricacies of semantic SEO, focusing on concepts like topical authority, the importance of understanding context in content creation, and the evolution of Google’s search algorithms away from PageRank towards a more semantic understanding. It emphasizes the need for a well-structured semantic content network that meets both macro and micro semantic needs to improve search rankings. Ultimately, it advocates for optimizing content not just for search engines, but with a deep understanding of human language and intent.
![Google didn’t give up.
“The destiny of [Google’s
search engine] is to become
that Star Trek computer, and
that’s what we are building.”
- Amit Singhal](https://image.slidesharecdn.com/seomasterysummit-saigon-filtered-230315194354-7adf39a3/85/The-Reason-Behind-Semantic-SEO-Why-does-Google-Avoid-the-Word-PageRank-120-320.jpg)
The Reason Behind Semantic SEO: Why does Google Avoid the Word PageRank?
- 1. The Reason Behind Semantic SEO Why does Google avoid the word “PageRank”
- 2. What’s up with Topical Authority? Am I against buying links? No....
- 3. What’s Up with Topical Authority? Do I Buy Links? Yes....
- 4. What’s Up with Topical Authority? Does SEO have to make sense? If you don’t read, No....
- 5. What’s Up with Topical Authority? RankMerge RankMerge = Actual Traffic + PageRank
- 6. What’s Up with Topical Authority? RankMerge Why is PageRank for a blind librarian?
- 7. What’s Up with Topical Authority? How did I structure the concept? Topical Authority was a protest, but not against link building.
- 8. What’s Up with Topical Authority? Topical Authority = Topical Coverage + Historical Data
- 9. Samples
- 10. Sample 1 – TheCoolist From 800,000 to over 3,000,000 clicks a month. Over 450,000 new ranking queries. Language: English Industry: Generic TheCoolist.com
- 11. Fill Topical Gaps Entertainment, Architecture, Social Skills –> Personality Types
- 12. Prioritize Topics Create Momentum Increase Ranking Priority for a Topic.
- 13. Design a Semantic Content Network Topical Map + Semantically Organized Content Network with a Knowledge Base
- 14. Nothing is Random in a Semantic Content Network Contextual Vector + Contextual Hierarchy + Contextual Structure + Contextual Connection + Contextual Coverage and Flow Every heading, paragraph, anchor text, list item, sentences before listings, sentences after listings, First sentence after heading, anchor texts, anchor text positions, tables, table columns, question and answer formats... Everything is about context in Semantics.
- 15. Sample 2 - Svalbardi We sell a botle of water for 90 Euroes. My favourite business model. A Shopify Store with 27 Articles with Semantics
- 16. Get Classified with Top Authorities Language: English Industry: Luxury Water Understand Macro Semantics (Topical Map + Context Distribution)
- 17. Topical Authority requires Topical Consolidation From B2B to D2C Transition. Taking Investment. Understand Micro Semantics (Semantic Content Network + Context Construction)
- 18. Create with Order. Unite with Momentum. An example of Topical Map and Semantic Content Network Creation
- 19. A Diagram for a Topical Map Only %5 is completed.
- 20. Understand Ranking Algorithms Always exceed the Quality Thresholds. (Initial and Re-ranking)
- 21. How to Take Authority of another website? One of the 23 SEO Case Study. • Get associated with the Top Authority Websites. • Create Site-wide Topical Entries and Context Vectors. • Get most of your traffic from the most important topic. • Outrank the top authorities to use Algorithmic Hierarchy. Optimize Macro and Micro semantics.
- 22. Practical Suggestions for Semantic SEO Vastness-Depth-Momentum
- 23. ‘X website outranks Y website for Z topic for P months’ Creates a Re-ranking Trigger for related topics.
- 25. Sample 3 - ABCFinance What is Ranking State?
- 26. Sample 4 - Roxiecosmetics From a Negative Ranking State to Positive One.
- 27. But When? After a Broad Core Algorithm Update.
- 28. Reflect Main Topic of Website
- 29. Where you get traffic determines main topic of website. Website A Has 999 pages about Porn Has 1 page about Bible Has 1 sessions for these 999 pages. Has 1 million organic sessions for this single page. Is this a Porn or Bible website?
- 30. Link most Quality Web pages from Homepage for Targeted Topics Understand Quality Nodes.
- 31. Link most Quality Web pages from Homepage for Targeted Topics
- 32. Macro and Micro Semantics 4 more Samples
- 33. Sample 5 - Nasdaq A Company from NASDAQ. – I don’t want to pay half million, sorry.
- 34. Sample 5 - Nasdaq I bought stocks since I do SEO... SEO doesn’t correlate with stock prices.
- 35. Sample 6 - Snuffstore Language: German - Industry: E-cigaratte – Semantics are Language Agnostic.
- 36. Sample 7 - Kanbanize Language: English - Industry: Task Management
- 37. Algorithmic Authorship and Content Engineering Art of organizing humans with algorithmic rules of writing, and content structuring according to LLMs.
- 38. Which sentence is more relevant? Query: ‘What is a penguin’ Penguin is a flightless seabirds with flippers instead of wings that live almost exclusively below the equator. Penguin is a flightless seabirds that live almost exclusively below the equator. Penguin is a flightless seabirds that live almost exclusively below the equator and they have flippers instead of wings. Sentence 1 Sentence 2 Sentence 3 Query: ‘Where does a Penguin Live?’
- 39. Prioritize Attributes and Contexts Interrogative Term: ‘Where’ is a signal for place. Penguins live below the equator, in the X, Y, Z geographies because their flippers and flightless Seabirds nature provide..... Query: ‘Where does a Penguin Live?’
- 40. Prioritize Attributes and Contexts Learn Query Processing, and Understanding. What happens if query is a single word Like, ‘Penguin’?
- 41. “Understanding” is needed. • Google focused on “Understanding”. • Microsoft Bing followed the same purpose, and created “Satori”.
- 42. “Understanding” is needed. • It is not about Stuffing Entities. • It is about Creating a Knowledge-base in the form of Content.
- 43. “Understanding” is needed. • Calculate Attribute Distance. • Understand Semantic Similarity. • Understand Semantic Relevance.
- 44. “Understanding” is needed. • Strength-based Re-ranking. • Coverage-based Re-ranking.
- 45. Instructions
- 46. Be Cheaper
- 47. Cost of Retrieval (Another concept for me) Source: Search Quality Meeting: Spelling for Long Queries (Annotated) – March 12, 2012
- 48. What is the Cost of Understanding you? Source: Improving Search Over the Years – 1 April 2020 Source: Webmaster Hangout, Zurich, 2021.
- 49. Cost of Ranking A Website can’t be higher than Cost of Not Ranking the Website.
- 50. Which one is Cheaper? Website B 1400 Content Item (Web Pages) 1200 Triples 94% Accuracy 7 Connected Topics Clarity %96 Website A 600 Content Item (Web Pages) 900 Triples 59% Accuracy 3 Connected Topics Clarity %78
- 51. To satisfy 9 Million Search Queries I won’t rank 9 million different website.
- 52. Topical Authority is born from the idea of Decreasing the Cost of Retrieval.
- 54. Optimize for Large Language Models, not for Blind Librarians Information Retrieval Information Extraction • Query to Document Relevance • Document to Query Relevance • Document to Document Similarity • Query to Query Similarity • Query to Question, and Question to Answer. • Focus on Question and Answer Generation for IR.
- 55. Generate Unique Questions Information Responsiveness Information Responsiveness is involving the respective entity, attribute, Value combinations with a macro and micro context distribution across multiple Semantically organized content network to prove overall quality and relevance of information.
- 56. Sample 9 – Entity Identity Creation Use Information Responsiveness, Quality and Accuracy to Change Google’s perception against Popular and High-level PageRank sources. If you can convince Google that ‘X is Y’ but not ‘Z’ with semantics, İt means that your topical map will be knowledge base. A website will rank only if they are similar to you. Make your competitors imitate you.
- 57. Sample 9 – Entity Identity Creation
- 59. Prepare for Future Generative AI Search
- 60. Sample 10 – AI Search Language: English Industry: Alcoholic Drinks
- 61. Sample 10 - AI Search Language: English Industry: Alcoholic Drinks
- 62. Sample 10 - AI Search Language: English Industry: Alcoholic Drinks
- 63. Sample 11 - Human Search Language: English Industry: Promised to not share
- 64. Sample 11 - Human Search Language: English Industry: Promised to not share
- 65. Sample 12 – Ultra AI Search
- 66. Sample 13 – Ultra AI Search
- 67. LLMo Optimization for SEO 1. Fine-tune a LLM. 2. Create a Topical Map. 3. Create a Semantic Content Network. 4. Generate Content 5. Include Human Effort 6. Improve your Knowledge Base 7. Make your website a Speaking AI. Human Effort with Microsemantic Optimization is must.
- 68. Find a Knowledge Base
- 69. Verbalize the Knowledge Base
- 70. Use Natural Language Inference with Chain of Reasoning
- 72. Optimize for Web Entity, not Website Web Entity: Website Social Media Accounts CEO Teammates Products Sub-brands Departments Local Address Environmental Policy Scholarship Website: ...
- 73. Who is gonna win?
- 74. The SEO who understands, not the SEO who imitates.
- 75. Semantics are Language Agnostic. Source: Facilitating communications with automated assistants in multiple languages
- 76. Cross-lingual Embeddings for Semantic Search A Cross Lingual Embeddings Example for the Sentence ‘Semantic search is an opportunity for Conversational Generative AI Search.’
- 77. Some Research Papers and Patents from Google Word Embeddings with Semantic Distance and Context Associations.
- 78. Some Research Papers and Patents from Google Word Embeddings with Semantic Distance and Context Associations.
- 79. Some Research Papers and Patents from Google Word Embeddings with Semantic Distance and Context Associations.
- 80. Some Research Papers and Patents from Google David C. Taylor – Context and Knowledge Domains with Embeddings
- 81. Some Research Papers and Patents from Google Hexagon is connected to Ultrasonic. Ultrasonic is not a primary connection to Hexagon.
- 82. Some Research Papers and Patents from Google «Hexagon» is an «Ultrasonic Wave» type.
- 83. Some Research Papers and Patents from Google ‘Quartz’ and ‘Cleaner’. Macro Context: ‘Industrial Ultrasonic Cleaning Machines‘.
- 84. How to Understand Language Models and Semantic Connections within them ‘Quartz’ and ‘Cleaner’.
- 85. Understand Search Engine ‘Cleaners’ and ‘ultrasonic’
- 86. How to Understand Language Models and Semantic Connections within them ‘Cleaner’ and ‘Ultrasonic’.
- 87. How to Understand Language Models and Semantic Connections within them ‘Quartz’ and ‘Ultrasonic’.
- 88. How to Understand Language Models and Semantic Connections within them ‘Quartz’ and ‘Ultrasonic’. • ‘Quartz’ and ‘Ultrasonic’. • Macro Context: ‘ Quartz Ultrasonic Absorption and Measurement‘. • Micro Context: Ultrasonic Cleaning for Qaurtz Surfaces • Zertec and its Products is signifier.
- 89. What is Large Language Model Optimization? A Language Interpretiy Tool by Google is published. You can check ‘Word Compositionality’ and ‘Embeddings’ for certain contexts.
- 90. How to Understand Language Models and Semantic Connections within them Large Language Model Optimization and Answer Engine Optimization are different technical expressions for Semantic Search Engine Optimization. Sequence Modeling (Word Compositionality Modeling) İs backbone of Semantic SEO. • What is the possibility of ‘Cat’ appears with predicates of ‘chase’, ‘eat’, or ‘fly’?. Optimize Sequences of Words. (Sequence Modeling).
- 91. How to Understand Language Models and Semantic Connections within them A manual exercise for ‘sensing search engine’
- 92. An Example of Relevance Configuration ‘Financial Advisor helps families to achieve financial independence.’ ‘Families achieve financial independence with the help of the financial advisor.’ Macro Context: ‘Financial Advisor’ Macro Context: ‘Family Economics’ Possible Search Queries: ‘Financial Advisor + Family’ Possible Representative Question: ‘What does Financial Advisor help families for?’ Possible Search Queries: ‘Family + Financial Independence’ Possible Representative Question: ‘How does a family achieve financial independence?’
- 93. How to Understand Language Models and Semantic Connections within them ‘Purple Yam’ -> ‘Sweet Potatoes’
- 94. What’s the game of Google?
- 95. How does Google Prepare itself?
- 96. From Document Ranking to Contextual Answer Ranking
- 97. From Document Ranking to Contextual Answer Ranking • Quality • Sensibleness • Specifity • Interestingness • Safety • Groundedness
- 98. From Document Ranking to Contextual Answer Ranking
- 99. From Document Ranking to Contextual Answer Ranking
- 101. Semantic SEO Course
- 102. Whatever-works-hat
- 103. ‘Black is the new white.’ Koray Tuğberk GÜBÜR
- 104. Thank you, Saigon! Did I write Saigon correctly?
- 105. There is no difference between keyword stuffing and entity stuffing. Gibberish is gibberish. What is a Blind Librarian?
- 106. Sources for Future AI Search with Conversations • Generation of text segment dependency analysis using neural networks • Facilitating communications with automated assistants in multiple languages • Processing techniques for text capture from a rendered document • Training and/or determining responsive actions for natural language input using coder models • Automatically determining language for speech recognition of spoken utterance received via an automated assistant interface • Non-deterministic task initiation with personal assistant module • Systems and methods for biomechanically-based eye signals for interacting with real and virtual objects • On-device projection neural networks for natural language understanding • End of query detection • Voice recognition system • Annotations in software applications for invoking dialog system functions • Enhancing functionalities of virtual assistants and dialog systems via plugin marketplace • Determining Dialog States for Language Models • Parameter collection and automatic dialog generation in dialog systems
- 107. Google merged with Oingo • Company started to focus on “Information Extraction”, not to “Information Retrieval” and “PageRank”. • Oingo is the inventor of Open Information Extraction. • Open Information Extraction is to create a structured data network from prose-type content. • It requires to create a “Knowledge Base”.
- 108. Sergey Brin tried first Semantic Search Engine attempt in 1999. • The First Semantic Search Engine Patent is filled in 2001. • It focused on “Patterns” and “Relations” in databases. • It didn’t work out due to high cost, low confidence.
- 109. Authors, Books, HTML Tags, URL Patterns, and Queries • Books and Authors (Tuples) were the first trial of the Semantic Search Engine creation. • But there were problems; fake authors, wrong titles, wrong genre names. Source: Extracting Patterns from Databases, Sergey Brin
- 110. DUAL Iterative Pattern Extraction became a norm. • Sergey Brin suggested using “Dual Iterative Pattern Extraction (DIPRE). • Dual means a “tuple” in the form of an “entity” and an “attribute”. • Pattern recognition became a fundamental step for semantic search.
- 111. Google didn’t give up. • They created “Phrase-posting” lists for different contexts. • They extracted “Co-occurrences” to construct a Proximity Search methodology. • They used tokenization, lemmatization, and stemming for words. • They have used TF-IDF, BM25, and Query Likelihood models. • They have invented Word2Vec, and GlovE. • But none of these were good enough to change the state from “Blind Librarian” to “Understanding Search Engine”.
- 112. Is TF-IDF so good? No… TF-IDF helps SEOs for many years to understand Document Statistics, but it has some flaws. • Longer Documents has “Higher Relevance” for TF-IDF. • Document Frequency (Corpus Size) manipulates the results. • Term Frequency for X is 50, Document Size is 50, Relevance 0. • Term Frequency for X is 50, Document Size 51, Relevance 1%. • Term Frequency for X is 50, Document Size 5001, Relevance is 90%+. • Term Frequency is 500 (Because document is longer), Document Size is 900, relevance will be increased further. We are still blind. Source: Christophher D. Manning, and Prabhakar Raghavan
- 113. Is B25 so good? No… A document that mentions “cat” 60 times is not twice more as relevant as a document that mentions “cat” 30 times. • BM25 has an extra parameter (k1) to normalize “term saturation”. How about short articles? • BM25 finds short articles more relevant. • Long-form articles lost relevance. We are still blind. Source: Christophher D. Manning, and Prabhakar Raghavan Term Saturation x Term Stuffing
- 114. An example of Lexical Search If query has both “USA” and “Moon”, the phrase posting list Moon -> 24, 66, 54, 21, 09, 43, 421 USA -> 42, 31, 56, 72, 31, 54, 51 • 54 exists on both inverted index. • “…. Bla bla bla bla….. Moon…… bla bla bla USA….. Bla bla bla….. • We are still blind.
- 115. Lastly, Query Likelihood … What is the possibility of ‘Query terms’ to appearing in a document? Query Likelihood is a transition between ‘Lexical Search’ and ‘Semantic Search’. But still, it doesn’t understand. Transition starts with LLMs.
- 116. “Understanding” is needed. • Google focused on “Understanding”. • Microsoft Bing followed the same purpose, and created “Satori”.
- 117. “Understanding” is needed. • It is not about Stuffing Entities. • It is about Creating a Knowledge-base in the form of Content.
- 118. “Understanding” is needed. • Calculate Attribute Distance. • Understand Semantic Similarity. • Understand Semantic Relevance.
- 119. “Understanding” is needed. • Strength-based Re-ranking. • Coverage-based Re-ranking.
- 120. Google didn’t give up. “The destiny of [Google’s search engine] is to become that Star Trek computer, and that’s what we are building.” - Amit Singhal
- 121. Google didn’t give up. “Google is designed for users, not for websites.” - Lawrance Page
- 122. How to Understand Language Models and Semantic Connections within them ‘Purple Yam’ -> ‘Sweet Potatoes’
- 123. How to Understand Language Models and Semantic Connections within them ‘Breville Juicer’ -> ‘Product’ + ‘Book’ Search
- 124. Some Research Papers and Patents from Google