Generative Engine Optimization (GEO): оптимизация под ChatGPT Search и Perplexity

[ LLM Search Engine (Perplexity) ]
       | (Query Vectorization -> Similarity Search)
       v
[ Vector Database / Search Index ] -> Retrieves Top-K Documents
       |
       v
[ Context Window ] -> [ Document A (High Density) ] + [ Document B (Low Density) ]
       | (LLM Synthesis & Citation)
       v
[ Final Answer ] -> "Согласно [1], архитектура строится на..."
       * Document A is cited (GEO optimized)
       * Document B is ignored

import { Metadata } from "next";
import { Article } from "@/types";

interface Props {
  params: Promise<{ slug: string }>;
}

export async function generateMetadata({ params }: Props): Promise<Metadata> {
  const { slug } = await params;
  return {
    title: "Understanding React Server Components",
    description: "Deep dive into RSC architecture and data fetching patterns.",
  };
}

export default async function BlogPostPage({ params }: Props) {
  const { slug } = await params;
  // Fetch article data...
  const article: Article = await fetchArticle(slug);

  // GEO Strategy: Strict JSON-LD for LLM Crawlers
  const jsonLd = {
    "@context": "https://schema.org",
    "@type": "TechArticle",
    "headline": article.title,
    "abstract": article.tldr, // Critical for LLM summaries
    "author": {
      "@type": "Person",
      "name": "Jane Doe",
      "url": "https://agenticlab.dev/authors/jane"
    },
    "about": {
      "@type": "SoftwareSourceCode",
      "programmingLanguage": "TypeScript",
      "codeRepository": "https://github.com/vercel/next.js"
    },
    // High semantic density statements for RAG indexing
    "mainEntity": {
      "@type": "ItemList",
      "itemListElement": article.keyFindings.map((finding, idx) => ({
        "@type": "ListItem",
        "position": idx + 1,
        "name": finding
      }))
    }
  };

  return (
    <article className="prose lg:prose-xl">
      <script
        type="application/ld+json"
        dangerouslySetInnerHTML={{ __html: JSON.stringify(jsonLd) }}
      />
      <h1>{article.title}</h1>
      
      {/* High density summary block optimized for LLMs */}
      <section className="tldr-block border-l-4 border-blue-500 pl-4 mb-8" data-nosnippet="false">
        <h2>Executive Summary</h2>
        <p>{article.tldr}</p>
      </section>

      <div dangerouslySetInnerHTML={{ __html: article.content }} />
    </article>
  );
}

// Mock function for compilation
async function fetchArticle(slug: string) {
  return { title: "Title", tldr: "TLDR", content: "<p>Content</p>", keyFindings: ["Fact 1", "Fact 2"] };
}