👁 Image
EasyAppDev.Blazor.AutoComplete.AI 1.0.7

EasyAppDev.Blazor.AutoComplete.AI

AI-powered semantic search for the Blazor AutoComplete component. Search by meaning, not just keywords.

👁 NuGet
👁 NuGet Downloads
👁 License: MIT

Features

• Semantic Search: Understands meaning, synonyms, and concepts
• SIMD Acceleration: 3-5x faster similarity calculations
• Dual Caching: Separate item/query caches with LRU eviction
• Multiple Providers: OpenAI, Azure OpenAI, Ollama, or custom
• Hybrid Search: Falls back to text matching for short queries

Installation

dotnet add package EasyAppDev.Blazor.AutoComplete.AI

Note: This package includes the core AutoComplete component as a dependency.

Quick Start

1. Configure OpenAI

Option A: appsettings.json (Recommended)

{
 "OpenAI": {
 "ApiKey": "sk-proj-...",
 "Model": "text-embedding-3-small"
 }
}

// Program.cs
builder.Services.AddAutoCompleteSemanticSearch(builder.Configuration);

Option B: Direct API Key

builder.Services.AddAutoCompleteSemanticSearch(
 apiKey: "sk-proj-...",
 model: "text-embedding-3-small");

2. Use the Component

@using EasyAppDev.Blazor.AutoComplete.AI

<SemanticAutoComplete TItem="Document"
 Items="@docs"
 SearchFields="@(d => new[] { d.Title, d.Description })"
 TextField="@(d => d.Title)"
 SimilarityThreshold="0.15"
 @bind-Value="@selectedDoc"
 Placeholder="Search by meaning..." />

Supported Providers

Azure OpenAI

builder.Services.AddAutoCompleteSemanticSearchWithAzure(
 endpoint: "https://my-resource.openai.azure.com/",
 apiKey: "your-azure-api-key",
 deploymentName: "text-embedding-ada-002");

Ollama (Local)

using OllamaSharp;

var ollamaClient = new OllamaApiClient("http://localhost:11434");
builder.Services.AddSingleton<IEmbeddingGenerator<string, Embedding<float>>>(
 ollamaClient.AsEmbeddingGenerator("nomic-embed-text"));

Advanced Configuration

<SemanticAutoComplete TItem="Product"
 Items="@products"
 SearchFields="@(p => new[] { p.Name, p.Description })"

 @* Cache Configuration *@
 ItemCacheDuration="TimeSpan.FromHours(2)"
 QueryCacheDuration="TimeSpan.FromMinutes(30)"
 MaxItemCacheSize="20000"

 @* Pre-warming *@
 PreWarmCache="true"
 ShowCacheStatus="true"

 @* Search Tuning *@
 SimilarityThreshold="0.15"
 MinSearchLength="3"
 DebounceMs="500"

 @bind-Value="@selectedProduct" />

Vector Database Providers

For production-grade semantic search with persistent storage, use vector database providers:

Performance

• SIMD Acceleration: 3-5x faster cosine similarity
• Cache Hit Rate: 80%+ after warm-up
• API Cost Reduction: 80%+ due to caching

License

MIT