{"slug":"koray-microsemantics","title":"Koray Microsemantics — Sentence and Passage Level Optimization","tags":["seo","semantic","koray","microsemantics","passage-ranking"],"agent_summary":"Optimizes individual sentences and paragraphs for passage ranking using Koray's microsemantics: predicate/noun optimization, answer format selection, distributional semantics, and word proximity.","trigger_phrases":["microsemantics","passage ranking","sentence optimization","answer format","Koray passage","word proximity SEO","distributional semantics","passage-level SEO","predicate optimization"],"runnable":true,"markdown":"\n## Core Concept: What Is Microsemantics?\n\nMicrosemantics is the optimization of meaning at the sentence and paragraph level — the smallest unit that Google evaluates for relevance during passage ranking. After 2021, Google began ranking individual passages, not just full documents. A page can rank for multiple queries if individual passages each fulfill different search intents.\n\nMicrosemantics controls:\n- Word order (predicate + noun placement)\n- Answer format selection (definition, comparison, procedure, value)\n- Word proximity (relevant terms must appear close together)\n- Discourse integration (sentences must connect logically)\n- Distributional relevance (context terms distributed across the passage, not front-loaded)\n\n## Predicate + Noun Optimization\n\nEvery sentence in main content should follow: **Subject → Predicate → Object** in that order, with the predicate defining the relationship between entities.\n\nWeak: \"Germany is one of the countries where work visas can be hard to get for some people.\"\nStrong: \"Germany requires non-EU citizens to obtain a work visa before beginning employment.\"\n\nThe strong version:\n- Puts the central entity (Germany) as subject\n- Uses a precise predicate (requires)\n- Specifies the object (non-EU citizens + work visa + employment)\n\n## The Four Answer Formats\n\nEach question type has an optimal answer format. Using the wrong format reduces passage ranking probability.\n\n| Question Type | Optimal Format | Example |\n|--------------|---------------|---------|\n| Definition (\"What is X?\") | Single declarative sentence, 25-50 words, starts with entity | \"A German work visa is a permit that authorizes non-EU nationals to live and work in Germany for a period exceeding 90 days.\" |\n| Comparison (\"X vs Y\") | Table or parallel structure with explicit differentiator stated first | Table: X | Y with clear column headers |\n| Procedure (\"How to X?\") | Numbered list, each step action-verb first | \"1. Gather required documents. 2. Book appointment...\" |\n| Value/Cost (\"How much X?\") | Numeric range in first sentence with units | \"German work visa fees range from €75 to €130 depending on visa type.\" |\n\n## Word Proximity Rules\n\nGoogle's term-proximity scoring rewards pages where related terms appear close together. Distributional semantics research shows that terms within 3-5 words of each other get stronger co-occurrence credit than terms 20+ words apart.\n\nRules:\n- Keep the entity and its core attribute within 5 words in the first sentence of each paragraph\n- Repeat key entity-attribute pairs in closing sentences of each section\n- Do not spread synonyms across a 500-word section — cluster them in the same paragraph\n\n## Discourse Integration\n\nDiscourse integration is the sentence-to-sentence logical flow that Google's language models evaluate when scoring passage coherence.\n\nStrong discourse integration:\n- Each sentence concludes a thought that the next sentence builds on\n- Transition words signal the logical relationship (however, therefore, as a result, by contrast)\n- Pronouns reference clear antecedents\n\nBroken discourse:\n- Sentences that could be shuffled without changing meaning (list-style without connectives)\n- Abrupt topic changes within a paragraph\n- Orphan sentences with no connection to surrounding content\n\n## Distributional Relevance\n\nContext terms must be distributed across the full passage, not front-loaded or back-loaded. A passage that contains all its relevant terms in the first 3 sentences and generic filler for the remaining 10 sentences has low distributional relevance.\n\nTarget: relevant context terms in every 3-sentence window throughout the passage.\n\n## Anti-Patterns\n\n- Passive voice as default (weakens predicate-object relationship)\n- Definition sentences longer than 60 words (exceed featured snippet extraction window)\n- Numbered lists where prose would create stronger discourse integration\n- Paragraph-opening sentences that restate the heading without adding information\n\n## Cross-References\n\n- `koray-contextual-vector` — heading-level structure that passages must support\n- `koray-content-audit` — auditing existing content for discourse integration violations\n- `featured-snippet-optimizer` — passage-level targeting for rich results\n\n#seo-sop #seo #semantic #koray #microsemantics #passage-ranking\n","html":"<h2>Core Concept: What Is Microsemantics?</h2>\n<p>Microsemantics is the optimization of meaning at the sentence and paragraph level — the smallest unit that Google evaluates for relevance during passage ranking. After 2021, Google began ranking individual passages, not just full documents. A page can rank for multiple queries if individual passages each fulfill different search intents.</p>\n<p>Microsemantics controls:</p>\n<ul>\n<li>Word order (predicate + noun placement)</li>\n<li>Answer format selection (definition, comparison, procedure, value)</li>\n<li>Word proximity (relevant terms must appear close together)</li>\n<li>Discourse integration (sentences must connect logically)</li>\n<li>Distributional relevance (context terms distributed across the passage, not front-loaded)</li>\n</ul>\n<h2>Predicate + Noun Optimization</h2>\n<p>Every sentence in main content should follow: <strong>Subject → Predicate → Object</strong> in that order, with the predicate defining the relationship between entities.</p>\n<p>Weak: \"Germany is one of the countries where work visas can be hard to get for some people.\"\nStrong: \"Germany requires non-EU citizens to obtain a work visa before beginning employment.\"</p>\n<p>The strong version:</p>\n<ul>\n<li>Puts the central entity (Germany) as subject</li>\n<li>Uses a precise predicate (requires)</li>\n<li>Specifies the object (non-EU citizens + work visa + employment)</li>\n</ul>\n<h2>The Four Answer Formats</h2>\n<p>Each question type has an optimal answer format. Using the wrong format reduces passage ranking probability.</p>\n<p>| Question Type | Optimal Format | Example |\n|--------------|---------------|---------|\n| Definition (\"What is X?\") | Single declarative sentence, 25-50 words, starts with entity | \"A German work visa is a permit that authorizes non-EU nationals to live and work in Germany for a period exceeding 90 days.\" |\n| Comparison (\"X vs Y\") | Table or parallel structure with explicit differentiator stated first | Table: X | Y with clear column headers |\n| Procedure (\"How to X?\") | Numbered list, each step action-verb first | \"1. Gather required documents. 2. Book appointment...\" |\n| Value/Cost (\"How much X?\") | Numeric range in first sentence with units | \"German work visa fees range from €75 to €130 depending on visa type.\" |</p>\n<h2>Word Proximity Rules</h2>\n<p>Google's term-proximity scoring rewards pages where related terms appear close together. Distributional semantics research shows that terms within 3-5 words of each other get stronger co-occurrence credit than terms 20+ words apart.</p>\n<p>Rules:</p>\n<ul>\n<li>Keep the entity and its core attribute within 5 words in the first sentence of each paragraph</li>\n<li>Repeat key entity-attribute pairs in closing sentences of each section</li>\n<li>Do not spread synonyms across a 500-word section — cluster them in the same paragraph</li>\n</ul>\n<h2>Discourse Integration</h2>\n<p>Discourse integration is the sentence-to-sentence logical flow that Google's language models evaluate when scoring passage coherence.</p>\n<p>Strong discourse integration:</p>\n<ul>\n<li>Each sentence concludes a thought that the next sentence builds on</li>\n<li>Transition words signal the logical relationship (however, therefore, as a result, by contrast)</li>\n<li>Pronouns reference clear antecedents</li>\n</ul>\n<p>Broken discourse:</p>\n<ul>\n<li>Sentences that could be shuffled without changing meaning (list-style without connectives)</li>\n<li>Abrupt topic changes within a paragraph</li>\n<li>Orphan sentences with no connection to surrounding content</li>\n</ul>\n<h2>Distributional Relevance</h2>\n<p>Context terms must be distributed across the full passage, not front-loaded or back-loaded. A passage that contains all its relevant terms in the first 3 sentences and generic filler for the remaining 10 sentences has low distributional relevance.</p>\n<p>Target: relevant context terms in every 3-sentence window throughout the passage.</p>\n<h2>Anti-Patterns</h2>\n<ul>\n<li>Passive voice as default (weakens predicate-object relationship)</li>\n<li>Definition sentences longer than 60 words (exceed featured snippet extraction window)</li>\n<li>Numbered lists where prose would create stronger discourse integration</li>\n<li>Paragraph-opening sentences that restate the heading without adding information</li>\n</ul>\n<h2>Cross-References</h2>\n<ul>\n<li><code>koray-contextual-vector</code> — heading-level structure that passages must support</li>\n<li><code>koray-content-audit</code> — auditing existing content for discourse integration violations</li>\n<li><code>featured-snippet-optimizer</code> — passage-level targeting for rich results</li>\n</ul>\n<p>#seo-sop #seo #semantic #koray #microsemantics #passage-ranking</p>\n"}