Semantic Kernel

This project is a Semantic Kernel Demo built using Microsoft Semantic Kernel, Ollama, and MongoDB Atlas. It integrates semantic search, LLM-based chat completion, and memory storage to provide personalized movie recommendations based on embedded movie descriptions.

Features

• Semantic search using OpenAI embeddings via Ollama
• Chat-based recommendation system using Llama3-Groq
• Memory storage for movie embeddings using VolatileMemoryStore
• Integration with MongoDB Atlas for retrieving and storing movie data
• Custom plugins for embedding text and searching information

Technologies Used

• Microsoft Semantic Kernel
• Ollama (Ollama API client for embeddings)
• MongoDB Atlas
• .NET 9
• C#

Setup and Installation

Prerequisites

• Install .NET 9
• Setup a MongoDB Atlas account and get the connection string
• Get Ollama API endpoint details

Configuration

1. Clone the repository:

   git clone https://github.com/your-username/your-repo.git
   cd your-repo

2. Open the project in Visual Studio or your preferred IDE.
3. Update the following environment variables or constants in Program.cs:
   • OllamaEndpoint: Your Ollama API endpoint
   • MongoDBAtlasConnectionString: Your MongoDB Atlas connection string
   • CollectionName: MongoDB collection storing movie data
4. Restore dependencies:

   dotnet restore

5. Build the project:

   dotnet build

Running the Application

To start the application, run:

   dotnet run

Usage

1. The system will retrieve movie data from MongoDB Atlas.
2. It will embed movie descriptions using the Ollama embedding model.
3. Users can enter queries in the console to receive movie recommendations.

Plugins

TextEmbeddingPlugin

• Provides semantic search for movies.
• Searches by genre, name, and type.
• Uses Semantic Kernel's ISemanticTextMemory.

Example Query

User > Recommend me a sci-fi movie with time travel.
Assistant > You might like "Interstellar". Plot: ...