【Semantic Kernel】RAG（检索增强生成）

RAG（Retrieval-Augmented Generation）是一种结合信息检索和生成模型的自然语言处理方法。它通过检索相关文档片段作为生成模型的上下文，提高生成文本的准确性和相关性。RAG广泛应用于问答系统、对话系统和文本摘要等领域，兼具高效性和灵活性。公司的客户机器人都特别适合。下面的案例是使用GPT的embedding来向量化相关信息，然后通过关键字检索，最后把这些信息，结合用户问题送给gpt-4o，得到一个相对友好的回复结果。

代码：

/// <summary>
/// RAG(Retrieval Augmented Generation)检索增强生成
/// </summary>
public static class RAGSample
{
    public static async Task Exec()
    {
        
#pragma warning disable SKEXP0001
#pragma warning disable SKEXP0010
#pragma warning disable SKEXP0050
        ISemanticTextMemory memory = new MemoryBuilder()
        .WithOpenAITextEmbeddingGeneration("text-embedding-ada-002", Config.OpenAiKey)
        .WithMemoryStore(new VolatileMemoryStore())//内存数据库，生产环境不能用
        .Build();
        #region 数据导入Memory
#pragma warning restore SKEXP0050
#pragma warning restore SKEXP0010
#pragma warning restore SKEXP0001
        var sampleData = new Dictionary<string, string>
        {
            ["info4"]
            = "3年级2班有56人",
            ["info5"]
            = "小明在3年级2班",
            ["info6"]
            = "小明在第二中学",
            ["info7"]
            = "第二中学有1500人"
        };
        var i = 0;
        foreach (var entry in sampleData)
        {
            await memory.SaveReferenceAsync(
                collection: "个人信息",
                externalSourceName: "GitHub",
                externalId: entry.Key,
                description: entry.Value,
                text: entry.Value);
        }
        #endregion

        var builder = Kernel.CreateBuilder()
            .AddOpenAIChatCompletion(modelId: Config.OpenAiChatModel, Config.OpenAiKey);

        var kernel = builder.Build();
        var chatHistory = new ChatHistory(); 
        var chat = kernel.GetRequiredService<IChatCompletionService>();
        while (true)
        {
            Console.ResetColor(); 
            Console.WriteLine("----------学生提问：----------"); 
            var ask = Console.ReadLine(); 
            chatHistory.Clear();
            chatHistory.AddSystemMessage("基于下面的信息回复问题："); 
            await foreach (var answer in memory.SearchAsync(collection: "个人信息", query: ask, limit: 3, minRelevanceScore: 0.65d, withEmbeddings: true))
            { 
                chatHistory.AddSystemMessage(answer.Metadata.Description); 
            }
            chatHistory.AddUserMessage(ask); 
            Console.WriteLine(); 
            Console.ForegroundColor = ConsoleColor.Green; 
            Console.WriteLine("==========讲师回答：=========="); 
            AuthorRole? role = AuthorRole.Assistant; 
            var contentBuilder = new StringBuilder(); 
            await foreach (var reply in chat.GetStreamingChatMessageContentsAsync(chatHistory)) 
            { 
                if (reply.Role.HasValue && role != reply.Role) 
                { 
                    role = reply.Role; 
                } 
                Console.Write(reply.Content);
                contentBuilder.Append(reply.Content); 
            }
            chatHistory.AddMessage(role.Value, contentBuilder.ToString()); 
            Console.WriteLine();
        }

    }
}