How to fetch meta description, title, keyword, category from content using AI in .net?

Anubhav Sharma 1016 13 Aug 2025

How to fetch meta description, title, keyword, category from content using AI in .net?

1 Answers

ICSM Computer 13 Aug 2025

If you want to automatically fetch meta description, title, keywords, and category from a given piece of content using AI in .NET, you basically need a pipeline that:

Takes your article/blog/content as input.
Sends it to an AI model (e.g., OpenAI GPT, Azure OpenAI, Hugging Face, etc.)
Parses the model's structured output into your .NET objects.

1. Example Prompt for AI Extraction

You can design a prompt like:

Extract the following fields from the given content:
- Title: A concise headline
- Meta description: Summary under 160 characters
- Keywords: 5–10 relevant SEO keywords
- Category: Broad topic classification

Return JSON only in this format:
{
  "title": "...",
  "metaDescription": "...",
  "keywords": ["...", "..."],
  "category": "..."
}

Content:
[YOUR CONTENT HERE]

2. .NET C# Example with OpenAI

Here’s a working example using Azure OpenAI (but you can adapt for OpenAI API directly):

using System;
using System.Net.Http;
using System.Net.Http.Headers;
using System.Text;
using System.Threading.Tasks;
using Newtonsoft.Json.Linq;

class Program
{
    static async Task Main()
    {
        string content = @"Your blog/article text here...";

        string prompt = $@"
Extract the following fields from the given content:
- Title: A concise headline
- Meta description: Summary under 160 characters
- Keywords: 5–10 relevant SEO keywords
- Category: Broad topic classification

Return JSON only in this format:
{{
  ""title"": """",
  ""metaDescription"": """",
  ""keywords"": [""...""],
  ""category"": """"
}}

Content:
{content}
";

        string endpoint = "https://YOUR_AZURE_OPENAI_ENDPOINT/openai/deployments/YOUR_MODEL/chat/completions?api-version=2024-05-01-preview";
        string apiKey = "YOUR_API_KEY";

        using (var client = new HttpClient())
        {
            client.DefaultRequestHeaders.Authorization = new AuthenticationHeaderValue("Bearer", apiKey);

            var requestBody = new
            {
                messages = new[]
                {
                    new { role = "system", content = "You are an assistant that extracts SEO metadata from content." },
                    new { role = "user", content = prompt }
                },
                matokens = 300,
                temperature = 0.3
            };

            var json = Newtonsoft.Json.JsonConvert.SerializeObject(requestBody);
            var response = await client.PostAsync(endpoint, new StringContent(json, Encoding.UTF8, "application/json"));

            string result = await response.Content.ReadAsStringAsync();
            var parsed = JObject.Parse(result);
            string jsonResponse = parsed["choices"]?[0]?["message"]?["content"]?.ToString();

            Console.WriteLine("Extracted Metadata:\n" + jsonResponse);
        }
    }
}

3. Why Use AI Instead of Regex?

AI understands context — it can create SEO-friendly titles and summaries, not just extract first <p> tags.
It can classify category (e.g., “Technology”, “Travel”, “Health”) even if the article never explicitly says it.
AI can choose optimal keywords for SEO rather than just picking frequent words.

4. If You Want to Avoid Paid APIs

Use Local LLMs with Ollama or LM Studio and call them from .NET.
Use Hugging Face models like facebook/bart-large-cnn for summaries or keyBERT for keyword extraction (via Python interop or ONNX in C#).

5. Possible Architecture

Content (HTML/Text)
       ↓
[Pre-processing] (strip HTML, clean text)
       ↓
[AI Call] (Prompt for title, meta, keywords, category)
       ↓
[Parse JSON Output]
       ↓
Save to DB / Show in UI

If you want, I can give you a ready-to-use .NET method that:

Takes raw HTML or text
Cleans it
Sends it to AI
Returns strongly-typed SEOData model with Title, MetaDescription, Keywords, Category.

Markdown for AI

A clean, structured version of this page for AI assistants and LLMs.

Open .md

forum