Group-based licensing is a mechanism to make it easier to assign and mange product licenses for large sets of user accounts. In this article, we discuss how to use Microsoft Graph PowerShell SDK cmdlets to manage group-based license assignments in a Microsoft 365 tenant. Assigning licenses to groups is very much like direct assignments, but some differences exist.

https://office365itpros.com/2024/11/04/group-based-licensing-sdk/

​ 
Group-based licensing is a mechanism to make it easier to assign and mange product licenses for large sets of user accounts. In this article, we discuss how to use Microsoft Graph PowerShell SDK cmdlets to manage group-based license assignments in a Microsoft 365 tenant. Assigning licenses to groups is very much like direct assignments, but some differences exist.
https://office365itpros.com/2024/11/04/group-based-licensing-sdk/  Read More

​

टाटा प्ले के खिलाफ शिकायत दर्ज करने के लिए, आप इन चरणों का पालन कर सकते हैं:198 टाटा प्ले ग्राहक सहायता से संपर्क करें:(0983✓122✓4914 ) एयरटेल वेबसाइट पर जाएँ या अपने मोबाइल डिवाइस पर टाटा प्ले खोलें।

​टाटा प्ले के खिलाफ शिकायत दर्ज करने के लिए, आप इन चरणों का पालन कर सकते हैं:198 टाटा प्ले ग्राहक सहायता से संपर्क करें:(0983✓122✓4914 ) एयरटेल वेबसाइट पर जाएँ या अपने मोबाइल डिवाइस पर टाटा प्ले खोलें।  Read More

​

I need some help here… I have an array of cell addresses as input, which I would like to transform into an array of hyperlinks. Like this: 

I tried HYPERLINK(A1#), but that doesn’t seem to work. Any suggestions?

​I need some help here… I have an array of cell addresses as input, which I would like to transform into an array of hyperlinks. Like this:   I tried HYPERLINK(A1#), but that doesn’t seem to work. Any suggestions?  Read More

​

फोनपे से पैसे काटने पर क्या करें  में ” संपर्क करें.((,+91-629O-348-172) या “इतिहास” सेक्शन में जाएं। रिले लेन-डेन चुनें। “वापस लें” या “वापस लेने के लिए भुगतान करें” विकल्प चुनें। अगर पुराना नहीं है अर्थात, तो अपने बैंक से संपर्क करें.

​फोनपे से पैसे काटने पर क्या करें  में ” संपर्क करें.((,+91-629O-348-172) या “इतिहास” सेक्शन में जाएं। रिले लेन-डेन चुनें। “वापस लें” या “वापस लेने के लिए भुगतान करें” विकल्प चुनें। अगर पुराना नहीं है अर्थात, तो अपने बैंक से संपर्क करें.  Read More

​

Hi Community,

How can I disable the annoying tooltips / hover text when working in Azure Synpase Studio. When navigating around the page, the tooltip will appear multiple times, getting in the way of actually clicking where I want to go.

Example below

​Hi Community, How can I disable the annoying tooltips / hover text when working in Azure Synpase Studio. When navigating around the page, the tooltip will appear multiple times, getting in the way of actually clicking where I want to go.Example below  Read More

​

क्या मोबिक्विक पैसा वापस करेगा?”ट्रांज़ेक्शन” (91-629O-348-172) या “इतिहास” सेक्शन में जाएं। रिले लेन- डेन चुनें। “वापस लें” या “वापस लेने के लिए भुगतान करें” विकल्प चुनें। यदि भुगतान नहीं मिलता है, तो अपने बैंक से संपर्क करें,,, 

​क्या मोबिक्विक पैसा वापस करेगा?”ट्रांज़ेक्शन” (91-629O-348-172) या “इतिहास” सेक्शन में जाएं। रिले लेन- डेन चुनें। “वापस लें” या “वापस लेने के लिए भुगतान करें” विकल्प चुनें। यदि भुगतान नहीं मिलता है, तो अपने बैंक से संपर्क करें,,,   Read More

​

Hello SQL Community,

I’m reaching out to confirm whether the specific issue described in KB 2700641 has been resolved in SQL Server 2019. The issue in question, which occurred in SQL Server 2008 and 2008 R2, involved an assertion error during BULK INSERT or BCP operations without the CHECK_CONSTRAINTS option enabled. Although our tables do not have constraints to check during insert operations, we had to use the CHECK_CONSTRAINTS option as a workaround in SQL Server 2008.

I have reviewed the release notes for SQL Server 2012 – 2019 but couldn’t find an explicit mention of this particular fix. Before we remove the workaround from our code, could anyone confirm if this issue has indeed been addressed in SQL Server 2019? Or is there any documentation available that confirms this bug was fixed in a specific SQL Server version?

Thank you for any insights or official references you could provide!

​Hello SQL Community,I’m reaching out to confirm whether the specific issue described in KB 2700641 has been resolved in SQL Server 2019. The issue in question, which occurred in SQL Server 2008 and 2008 R2, involved an assertion error during BULK INSERT or BCP operations without the CHECK_CONSTRAINTS option enabled. Although our tables do not have constraints to check during insert operations, we had to use the CHECK_CONSTRAINTS option as a workaround in SQL Server 2008.I have reviewed the release notes for SQL Server 2012 – 2019 but couldn’t find an explicit mention of this particular fix. Before we remove the workaround from our code, could anyone confirm if this issue has indeed been addressed in SQL Server 2019? Or is there any documentation available that confirms this bug was fixed in a specific SQL Server version?Thank you for any insights or official references you could provide!  Read More

​

In the competitive landscape of e-commerce, delivering precise and relevant search results is crucial for enhancing user experience and driving sales. Traditional keyword-based search methods often struggle to grasp the nuanced intentions behind user queries. This is where vector similarity search comes into play, leveraging advanced embedding techniques to provide more accurate and context-aware search results. In this blog, we’ll walk you through implementing vector similarity search in an e-commerce products API using Azure Cosmos DB for NoSQL and TypeScript (JavaScript).

Introduction to Vector Similarity Search

Vector similarity search is a powerful technique that transforms data into high-dimensional vectors (embeddings) and enables the comparison of these vectors to find similarities. In the context of e-commerce, this allows for more intelligent product searches, recommendations, and categorization by understanding the semantic meaning behind product descriptions, tags, and features.

Embeddings

Modern machine learning models can be trained to convert raw data into embeddings, represented as arrays (or vectors) of floating-point numbers of fixed dimensionality. What makes embeddings useful in practice is that the position of an embedding in vector space captures some of the semantics of the data, depending on the type of model and how it was trained. Points that are close to each other in vector space are considered similar (or appear in similar contexts), and points that are far away are considered dissimilar.

Large datasets of multi-modal data (text, audio, images, etc.) can be converted into embeddings with the appropriate model. Projecting the vectors’ principal components in 2D space results in groups of vectors that represent similar concepts clustering together, as shown below.

Indexes

Embeddings for a given dataset are made searchable via an index. The index is constructed by using data structures that store the embeddings such that it’s very efficient to perform scans and lookups on them

Brute force search

The simplest way to perform vector search is to perform a brute force search, without an index, where the distance between the query vector and all the vectors in the database are computed, with the top-k closest vectors returned. This is equivalent to a k-nearest neighbours (kNN) search in vector space.

As you can imagine, the brute force approach is not scalable for datasets larger than a few hundred thousand vectors, as the latency of the search grows linearly with the size of the dataset. This is where approximate nearest Neighbour (ANN) algorithms come in.

Overview of Azure Cosmos DB Vector Search

Azure Cosmos DB for NoSQL now offers a vector search feature in preview, designed to handle high-dimensional vectors efficiently and accurately at any scale. This feature allows you to store vectors directly within your documents alongside traditional schema-free data. By colocating data and vectors, Azure Cosmos DB ensures efficient indexing and searching, simplifying data management and enhancing AI application architectures.

Key capabilities include:

• Vector Embedding Policies: Define how vectors are generated and stored within your documents.
• Vector Indexing: Optimize data retrieval based on vector similarities using various indexing methods.
• Vector Distance Functions: Utilize metrics like Cosine, Dot Product, and Euclidean to measure similarity.

By integrating vector similarity search into your e-commerce API, you can significantly enhance the relevance and personalization of search results.

What is a Vector Store?

A vector store or vector database is designed to store and manage vector embeddings—mathematical representations of data in a high-dimensional space. Each dimension corresponds to a feature of the data, and a vector’s position represents its characteristics. This enables efficient similarity comparisons for various data types, including text, images, audio, and more.

How Does a Vector Store Work?

Vector stores use vector search algorithms to index and query embeddings. These algorithms help find similar items based on data characteristics rather than exact matches. This technique is invaluable for applications like searching similar text, finding related images, making recommendations, and detecting anomalies. Vector search measures the distance between data vectors and a query vector, retrieving the most semantically similar results.

Azure Cosmos DB’s integrated vector database allows embeddings to be stored, indexed, and queried alongside original data, eliminating the need for separate vector databases and enhancing data consistency, scalability, and performance.

Prerequisites

Before diving into the implementation, ensure you have the following:

• Azure Cosmos DB for NoSQL Account: If you don’t have one, read how to create one here or use an existing Azure subscription.
• Azure Cosmos DB JavaScript SDK: Version 4.1.0 or later.
• Node.js and TypeScript: Familiarity with these technologies is essential.
• Vector Search Preview Registration: Required to access vector search features. Read more here
• Azure Account: Try Azure for free, if student Try azure for Students

Registering for Azure Cosmos DB Vector Search Preview

Vector search in Azure Cosmos DB is currently in preview and requires explicit registration. Follow these steps to enroll:

1. Navigate to Azure Portal:

   • Search for Azure Cosmos DB

1. • Click create
   • Create Azure Cosmos DB for NoSQL

1. • Please complete the process of provisioning, if you have any issue check this blog
   • go to your Azure Cosmos DB for NoSQL resource page.

2. Access Features Pane:

   • Select the “Features” pane under the “Settings” menu.

3. Enable Vector Search:

   • Find “Vector Search in Azure Cosmos DB for NoSQL.”
   • Read the feature description to ensure it meets your requirements.
   • Click “Enable” to enroll in the preview. After it is enabled, it should look like the below image

Note: The vector search feature is only supported on new containers. Existing containers cannot be retrofitted with vector search capabilities.

The registration request will be auto-approved, but it may take several minutes to take effect.

Keys we shall need in our env file

• Create and complete the provisioning process.
• After you’re done, go to overview and click Explore Azure AI Studio

• This will redirect you to Azure AI Studio, sign in with the same email account you signed in Azure Portal
• In the Azure AI Studio, Click Deployments, select Deploy base model from Deploy model.

• Select and check embeddings from Inference tasks menu.

• Select text-embedding-3-large and confirm to provision the model.
• Give your model a deployment name or go with the default and click deploy.
• After the deployment is complete copy the Target URI this will be the value of AZURE_OPENAI_ENDPOINT and Key will be the value of AZURE_OPENAI_API_KEY in our .env file.
   

• With that you should have below values

AZURE_COSMOS_DB_ENDPOINT=
AZURE_COSMOS_DB_KEY=
AZURE_COSMOS_DB=productsdb

AZURE_OPENAI_API_KEY=
AZURE_OPENAI_TEXT_EMBEDDING_MODEL_ENDPOINT=

Setting Up Azure Cosmos DB Container for Vector Search

Creating the Database and Container

Assume we’re building an internet-based store for smart home devices. Each product has properties like Name, Brand, SKU, Category, and Description. We’ll also define vector embeddings for descriptionVector, tagsVector, and featuresVector to enable vector-based searches.

Example JSON Structure:

{
  "name": "Smart LED Light Bulb",
  "brand": "BrightHome",
  "sku": "SKU-1002",
  "category": "Home Automation",
  "price": 19.99,
  "currency": "USD",
  "stock": 500,
  "description": "Illuminate your home with BrightHome Smart LED Light Bulbs. Control brightness and color temperature via smartphone or voice commands.",
  "features": "Wi-Fi Enabled, Voice Control Compatible, Energy Efficient, Adjustable Brightness, Long Lifespan, Easy Installation",
  "rating": 4.3,
  "reviewsCount": 850,
  "tags": [
    "lighting",
    "smart home",
    "LED",
    "energy-efficient",
    "voice-control"
  ],
  "imageUrl": "https://example.com/images/products/SKU-1002.jpg",
  "manufacturer": "BrightHome Technologies",
  "model": "BH-SMART-LB",
  "releaseDate": "2023-07-20",
  "warranty": "1 year",
  "dimensions": {
    "weight": "75g",
    "width": "6.5cm",
    "height": "11cm",
    "depth": "6.5cm"
  },
  "color": "White",
  "material": "Glass and Plastic",
  "origin": "Germany",
  "id": "a2555d74-686a-4c34-af2b-c63fa2b9ed2f"
}

Defining Vector Embedding Policy

The vector embedding policy informs the Cosmos DB query engine how to handle vector properties. It specifies the path, data type, dimensions, and distance function for each vector.

Typescript Example:

const vectorEmbeddingPolicy = {
  vectorEmbeddings: [
    {
      path: "/descriptionVector",
      dataType: "float32",
      dimensions: 1536,
      distanceFunction: "cosine",
    },
    {
      path: "/tagsVector",
      dataType: "float32",
      dimensions: 1536,
      distanceFunction: "cosine",
    },
    {
      path: "/featuresVector",
      dataType: "float32",
      dimensions: 1536,
      distanceFunction: "cosine",
    },
  ],
};

Adding Vector Indexes to the Indexing Policy

Vector indexes optimize the storage and retrieval of vector data. You must define these indexes during container creation.

Typescript Example:

const indexingPolicy = {
  vectorIndexes: [
    { path: "/descriptionVector", type: "diskANN" },
    { path: "/tagsVector", type: "diskANN" },
    { path: "/featuresVector", type: "diskANN" },
  ],
  includedPaths: [
    {
      path: "/*",
    },
  ],
  excludedPaths: [
    {
      path: "/descriptionVector/*",
    },
    {
      path: "/tagsVector/*",
    },
    {
      path: "/featuresVector/*",
    },
  ],
};

Note: During early preview, vector indexes can’t be modified once created. If changes are needed, create a new container with the updated vector index policy.

Creating the Container

With the embedding and indexing policies defined, create the container:

Typescript Example:

const containerName = "vector-embedding-container";

// Create container
const { resource: containerdef } = await database.containers.createIfNotExists({
  id: containerName,
  vectorEmbeddingPolicy: vectorEmbeddingPolicy,
  indexingPolicy: indexingPolicy,
});

Important: Vector search is only supported on new containers. Ensure both the vector embedding policy and vector indexing policy are set during container creation.

Building the Products API with Vector Search

We’ll build a TypeScript-based API using Express that interacts with Azure Cosmos DB to perform vector similarity searches.

Folder Structure

Project Setup

package.json:

{
  "name": "products-api-cosmosdb-typescript-vector-search",
  "version": "1.0.0",
  "description": "a Products API using Express and TypeScript, and integrate Azure Cosmos DB (with vector search capabilities) for managing product data. We'll also implement vector search, enabling us to query items based on vector similarity.",
  "scripts": {
    "dev": "tsx watch src/server.ts",
    "build": "rimraf dist && tsc",
    "start": "node dist/server.js"
  },
  "keywords": [],
  "author": "",
  "license": "ISC",
  "devDependencies": {
    "@types/cors": "^2.8.17",
    "@types/express": "^5.0.0",
    "@types/morgan": "^1.9.9",
    "@types/multer": "^1.4.12",
    "@types/node": "^22.7.9",
    "@types/uuid": "^10.0.0",
    "rimraf": "^6.0.1",
    "tsx": "^4.19.1",
    "typescript": "^5.6.3"
  },
  "dependencies": {
    "@azure/cosmos": "^4.1.1",
    "axios": "^1.7.7",
    "cors": "^2.8.5",
    "dotenv": "^16.4.5",
    "express": "^4.21.1",
    "express-rate-limit": "^7.4.1",
    "helmet": "^8.0.0",
    "morgan": "^1.10.0",
    "multer": "1.4.5-lts.1",
    "pino": "^9.5.0",
    "pino-pretty": "^11.3.0",
    "rotating-file-stream": "^3.2.5",
    "uuid": "^10.0.0",
    "zod": "^3.23.8"
  }
}

tsconfig.json:

{
  "compilerOptions": {
    /* Language and Environment */
    "target": "ES2022",
    /* Modules */
    "module": "CommonJS",
    "rootDir": "./src",
    "outDir": "./dist",
    "moduleResolution": "node",
    /* Interop Constraints */
    "esModuleInterop": true,
    "forceConsistentCasingInFileNames": true,
    "allowSyntheticDefaultImports": true,
    /* Type Checking */
    "strict": true,
    /* Completeness */
    "skipLibCheck": true
  },
  "exclude": [
    "node_modules",
    "dist"
  ],
  "include": [
    "src/**/*.ts"
  ]
}

Configuring Environment Variables

Create a .env file to manage sensitive configurations:

.env:

# Azure Cosmos DB Configuration
AZURE_COSMOS_DB_ENDPOINT=https://<your-cosmosdb-account>.documents.azure.com:443/
AZURE_COSMOS_DB_KEY=<your-cosmosdb-key>
AZURE_COSMOS_DB=<your-database-name>

# Azure OpenAI Configuration
AZURE_OPENAI_API_KEY=<your-openai-api-key>

# Azure OpenAI Text Embedding Model Configuration
AZURE_OPENAI_TEXT_EMBEDDING_MODEL_ENDPOINT=https://<your-openai-endpoint>.openai.azure.com/openai/deployments/<your-embedding-model-deployment>/embeddings?api-version=2023-05-15

Ensure you replace placeholder values with your actual credentials.

Setting Up the Server

src/server.ts:

//server/src/server.ts
import config from './config/env';
import logger from './utils/logger';
import CosmosDB from './services/cosmosdbService';
import initializeServices from './services/index';

import app from './app';



// Start server
const startServer = async () => {
    try {
        // Initialize CosmosDB
        const CosmosInstance = CosmosDB.getInstance();
        await CosmosInstance.initialize();

        // Initialize services after CosmosDB is initialized
        const services = initializeServices();
        
        // Attach services to app.locals for access in controllers
        app.locals.services = services;

        const PORT = config.port;

        app.listen(PORT, () => {
            logger.info(`Server is running on port ${PORT}`);
        });
    } catch (error) {
        logger.error('Failed to initialize Cosmos DB:', error);
        process.exit(1);
    }
};

startServer();

Implementing Logging

src/utils/logger.ts:

// src/utils/logger.ts
import pino from 'pino'

const logger = pino({
    base: {
        pid: false
    },
    transport: {
        target: 'pino-pretty',
        options: {
            colorize: true
        }
    }
})

export default logger;

Using Pino for efficient and structured logging.

Defining Routes

src/routes/productsRoutes.ts:

//src/routes/productsRoutes.ts
import express, { Request, Response } from "express";
import {
    createProduct,
    getProductById,
    getProducts,
    searchProductsByDescriptionVector,
    searchProductsByFeaturesVector,
    searchProductsByTagsVector
} from "../controllers/productController";

const productRouter = express.Router();


// Create a new product
productRouter.post('/products', (req: Request, res: Response) => {
    createProduct(req, res);
});

// Get all products
productRouter.get('/products', (req: Request, res: Response) => {
    getProducts(req, res);
});

// Get a product by ID
productRouter.get('/products/:id', (req: Request, res: Response) => {
    getProductById(req, res);
});

// VECTOR SEARCH ROUTES

// Search products by description vector
productRouter.post('/products/search/description', (req: Request, res: Response) => {
    searchProductsByDescriptionVector(req, res);
});

// Search products by tags vector
productRouter.post('/products/search/tags', (req: Request, res: Response) => {
    searchProductsByTagsVector(req, res);
});

// Search products by features vector
productRouter.post('/products/search/features', (req: Request, res: Response) => {
    searchProductsByFeaturesVector(req, res);
});



export default productRouter;

Implementing Controllers and Services

src/controllers/productController.ts:

// src/controllers/productController.ts
import { v4 as uuidv4 } from 'uuid';
import { Request, Response } from 'express';
import { productSchema, Product, productIDSchema, featuresSchema, tagsSchema } from '../schemas/productSchema';
import { generateTextVector } from '../services/embeddingService';
import logger from '../utils/logger';


/**
 * 
 *  req The request object.
 *  res The response object to send back the fetched products.
 * @returns The list of products.
 */
export const getProducts = async (req: Request, res: Response) => {
    // Access productService from app.locals
    const { productService } = req.app.locals.services;

    try {
        const products = await productService.getProducts();
        return res.status(200).json({
            status: 'success',
            data: products,
        });
    } catch (error: any) {
        logger.error('Error getting products:', error.message);
        return res.status(500).json({
            status: 'error',
            message: 'Internal Server Error',
        });
    }
};
/**
 * 
 *  req The request object containing product data. ie product id
 *  res The response object to send back the fetched product.
 * @returns The product associated with the given id.
 */
export const getProductById = async (req: Request, res: Response) => {
    // Access productService from app.locals
    const { productService } = req.app.locals.services;
    const { id } = req.params;
    
    // use zod to validate the id
    const parseIDResult = productIDSchema.safeParse({ id });
    if (!parseIDResult.success) {
        return res.status(400).json({
            status: 'error',
            message: 'Invalid product ID, should be a valid UUID',
        });
    }

    try {
        const product = await productService.getProductById(parseIDResult.data.id);

        if (!product) {
            return res.status(404).json({
                status: 'error',
                message: 'Product not found',
            });
        }

        return res.status(200).json({
            status: 'success',
            data: product,
        });
    } catch (error: any) {
        logger.error('Error getting product:', error.message);
        return res.status(500).json({
            status: 'error',
            message: 'Internal Server Error',
        });
    }
};

/**
 * Creates a new product with vector embeddings.
 *  req The request object containing product data.
 *  res The response object to send back the created product.
 */
export const createProduct = async (req: Request, res: Response) => {
    const { productService } = req.app.locals.services;

    // zod validation for a new product
    const parseResult = productSchema.safeParse(req.body);
    if (!parseResult.success) {
        const errors = parseResult.error.errors.map(err => ({
            path: err.path.join('.'),
            message: err.message,
        }));

        return res.status(400).json({
            status: 'error',
            errors,
        });
    }

    const product: Product = parseResult.data;

    try {
        if (!product.id) {
            product.id = uuidv4();
        }

        // Generate description vector
        if (product.description) {
            product.descriptionVector = await generateTextVector(product.description);
        }

        // Generate tags vector
        if (product.tags && product.tags.length > 0) {
            const tagsText = product.tags.join(' ');
            product.tagsVector = await generateTextVector(tagsText);
        }

        // Generate features vector
        if (product.features) {
            product.featuresVector = await generateTextVector(product.features);
        }

        const createdProduct = await productService.addProduct(product);

        return res.status(201).json({
            status: 'success',
            data: createdProduct,
        });
    } catch (error: any) {
        logger.error('Error creating product:', error.message);
        return res.status(500).json({
            status: 'error',
            message: 'Internal Server Error',
        });
    }
};

/**
 * Searches for products based on a tags vector.
 *  req The request object containing the query vector.
 *  res The response object to send back the search results.
 */
export const searchProductsByTagsVector = async (req: Request, res: Response) => {
    const { productService } = req.app.locals.services;
    const { queryTags, top } = req.body;
    // zod validation
    const parsedTags = tagsSchema.safeParse({ tags: queryTags });
    if (!parsedTags.data) {
        return res.status(400).json({
            status: 'error',
            message: 'Invalid tags. It should be an array of strings ie "queryVector" : ["item","item2"].',
        });
    }

    const tagsText = parsedTags.data.tags.join(' ');  // Convert the tags array to a single string
    const tagsVector = await generateTextVector(tagsText); // Generate the tags vector

    const topResults = top && Number.isInteger(top) && top > 0 ? top : 10;  // Get the top results ie 10 results

    try {
        const results = await productService.searchProductsByTagsVector(tagsVector, topResults);
        return res.status(200).json({
            status: 'success',
            data: results,
        });
    } catch (error: any) {
        logger.error('Error searching products by tags:', error.message);
        return res.status(500).json({
            status: 'error',
            message: 'Internal Server Error',
        });
    }
};

/**
 * Searches for products based on a features vector.
 *  req The request object containing the query vector.
 *  res The response object to send back the search results.
 */
export const searchProductsByFeaturesVector = async (req: Request, res: Response) => {
    const { productService } = req.app.locals.services;
    const { queryFeatures, top } = req.body;
    // zod validation
    const parsedFeature = featuresSchema.safeParse({ tags: queryFeatures });
    if (!parsedFeature.success) {
        return res.status(400).json({
            status: 'error',
            message: 'Invalid features. It should be an array of strings ie "queryVector" : ["item","item2"].',
        });
    }

    const tagsText = parsedFeature.data.tags.join(' ');
    const featuresVector = await generateTextVector(tagsText);

    const topResults = top && Number.isInteger(top) && top > 0 ? top : 10;

    try {
        const results = await productService.searchProductsByFeaturesVector(featuresVector, topResults);
        return res.status(200).json({
            status: 'success',
            data: results,
        });
    } catch (error: any) {
        logger.error('Error searching products by features:', error.message);
        return res.status(500).json({
            status: 'error',
            message: 'Internal Server Error',
        });
    }
};

/**
 * Searches for products based on a description vector.
 *  req The request object containing the query vector.
 *  res The response object to send back the search results.
 */

export const searchProductsByDescriptionVector = async (req: Request, res: Response) => {
    const { productService } = req.app.locals.services;
    const { queryDescription, top } = req.body;

    const descriptionVector = await generateTextVector(queryDescription);

    const topResults = top && Number.isInteger(top) && top > 0 ? top : 10;

    try {
        const results = await productService.searchProductsByDescriptionVector(descriptionVector, topResults);
        return res.status(200).json({
            status: 'success',
            data: results,
        });
    } catch (error: any) {
        logger.error('Error searching products by description:', error.message);
        return res.status(500).json({
            status: 'error',
            message: 'Internal Server Error',
        });
    }
};

src/services/productService.ts:

// server/src/services/productService.ts

import { Container } from '@azure/cosmos';
import CosmosDB from '../services/cosmosdbService';
import { Product } from '../schemas/productSchema';

class ProductService {
    private container: Container;

    constructor() {
        const cosmosDB = CosmosDB.getInstance();
        this.container = cosmosDB.getProductsContainer();
    }

    /**
     * Adds a new product to the Cosmos DB container.
     *  product The validated product to add.
     * @returns The created product resource.
     */
    public async addProduct(product: Product) {
        const { resource } = await this.container.items.create<Product>(product);
        return resource;
    }

    /**
     * Retrieves a product by its ID.
     *  id The ID of the product.
     * @returns The retrieved product or null if not found.
     */
    public async getProductById(id: string) {
        try {
            const { resource } = await this.container.item(id, id).read<Product>();
            return resource;
        } catch (error: any) {
            if (error.code === 404) {
                return null;
            }
            throw error;
        }
    }

    /**
     * Retrieves all products from the Cosmos DB container.
     * @returns An array of products.
     */
    public async getProducts() {
        const { resources } = await this.container.items
            .query("SELECT c.id, c.name, c.brand, c.sku, c.category, c.price, c.currency, c.stock, c.description, c.features, c.rating, c.reviewsCount, c.tags, c.imageUrl, c.manufacturer, c.model, c.releaseDate, c.warranty, c.dimensions, c.color, c.material, c.origin FROM c")
            .fetchAll();
        return resources;
    }

    /**
     * Searches for products similar to the provided description vector.
     *  queryVector The vector representing the search query.
     *  top The number of top similar products to retrieve.
     * @returns An array of products with similarity scores.
     */

    public async searchProductsByDescriptionVector(queryVector: number[], top: number = 10) {
        const querySpec = {
            query: `
                SELECT TOP @top 
                c.id, 
                    c.name, 
                    c.brand, 
                    c.sku, 
                    c.category, 
                    c.price, 
                    c.currency, 
                    c.stock, 
                    c.description, 
                    c.features, 
                    c.rating, 
                    c.reviewsCount, 
                    c.tags, 
                    c.imageUrl, 
                    c.manufacturer, 
                    c.model, 
                    c.releaseDate, 
                    c.warranty, 
                    c.dimensions, 
                    c.color, 
                    c.material, 
                    c.origin, 
                VectorDistance(c.descriptionVector, @queryVector) AS SimilarityScore
                FROM c
                ORDER BY VectorDistance(c.descriptionVector, @queryVector)
                `,
            parameters: [
                { name: "@queryVector", value: queryVector },
                { name: "@top", value: top }
            ]
        };

        const { resources } = await this.container.items.query(querySpec).fetchAll();
        return resources;
    }

    /**
     * Searches for products similar to the provided tags vector.
     *  queryVector The vector representing the tags search query.
     *  top The number of top similar products to retrieve.
     * @returns An array of products with similarity scores.
     */
    public async searchProductsByTagsVector(queryVector: number[], top: number = 10) {
        const querySpec = {
            query: `
                SELECT TOP @top 
                    c.id, 
                    c.name, 
                    c.brand, 
                    c.sku, 
                    c.category, 
                    c.price, 
                    c.currency, 
                    c.stock, 
                    c.description, 
                    c.features, 
                    c.rating, 
                    c.reviewsCount, 
                    c.tags, 
                    c.imageUrl, 
                    c.manufacturer, 
                    c.model, 
                    c.releaseDate, 
                    c.warranty, 
                    c.dimensions, 
                    c.color, 
                    c.material, 
                    c.origin, 
                    VectorDistance(c.tagsVector, @queryVector) AS SimilarityScore
                FROM c
                ORDER BY VectorDistance(c.tagsVector, @queryVector)
            `,
            parameters: [
                { name: "@queryVector", value: queryVector },
                { name: "@top", value: top }
            ]
        };

        const { resources } = await this.container.items.query(querySpec).fetchAll();
        return resources;
    }

    /**
     * Searches for products similar to the provided features vector.
     *  queryVector The vector representing the features search query.
     *  top The number of top similar products to retrieve.
     * @returns An array of products with similarity scores.
     */
    public async searchProductsByFeaturesVector(queryVector: number[], top: number = 10) {
        const querySpec = {
            query: `
                SELECT TOP @top 
                    c.id, 
                    c.name, 
                    c.brand, 
                    c.sku, 
                    c.category, 
                    c.price, 
                    c.currency, 
                    c.stock, 
                    c.description, 
                    c.features, 
                    c.rating, 
                    c.reviewsCount, 
                    c.tags, 
                    c.imageUrl, 
                    c.manufacturer, 
                    c.model, 
                    c.releaseDate, 
                    c.warranty, 
                    c.dimensions, 
                    c.color, 
                    c.material, 
                    c.origin, 
                    VectorDistance(c.featuresVector, @queryVector) AS SimilarityScore
                FROM c
                ORDER BY VectorDistance(c.featuresVector, @queryVector)
            `,
            parameters: [
                { name: "@queryVector", value: queryVector },
                { name: "@top", value: top }
            ]
        };

        const { resources } = await this.container.items.query(querySpec).fetchAll();
        return resources;
    }
}

export default ProductService;

src/services/index.ts:

// server/src/services/index.ts

import ProductService from './productService';
import CosmosDB from './cosmosdbService';

// Ensure CosmosDB is initialized before creating services
const initializeServices = () => {
    const cosmosInstance = CosmosDB.getInstance();
    // You can perform any additional service initialization here if needed
    return {
        productService: new ProductService(),
        // Add other services here
    };
};

export default initializeServices;

Generating Vector Embeddings

To perform vector similarity searches, we need to generate embeddings for product descriptions, tags, and features. We’ll use Azure OpenAI’s Embedding API for this purpose.

src/services/embeddingService.ts:

//src/services/embeddingService.ts
import axios from 'axios';
import config from '../config/env';

/**
 * Generates a vector embedding for a given text using OpenAI's Embedding API.
 *  text The input text to embed.
 * @returns An array of numbers representing the embedding vector.
 */
export async function generateTextVector(text: string): Promise<number[]> {
    const response = await axios.post(config.azureOpenAi.models.embedding.endPoint,
        { input: text },
        {
            headers: {
                'Content-Type': 'application/json',
                'api-key': config.azureOpenAi.apiKey,
            }
        });

    const vector = response.data.data[0].embedding;
    return vector;
}

This service takes input text and returns its vector embedding, which is then stored in Cosmos DB.

Initializing CosmosDB Service

src/services/cosmosdbService.ts:

//src/services/cosmosdbService.ts
import {
    PartitionKeyDefinitionVersion,
    PartitionKeyKind,
    CosmosClient,
    Container,
    CosmosDbDiagnosticLevel,
    VectorEmbeddingPolicy,
    IndexingPolicy,
    VectorIndexType,
    VectorEmbeddingDataType,
    VectorEmbeddingDistanceFunction
} from '@azure/cosmos';
import logger from '../utils/logger';
import config from '../config/env';

class CosmosDB {
    private static instance: CosmosDB;
    private client: CosmosClient;
    private productsContainer!: Container;
    private initialized: boolean = false;
    private initializing: Promise<void> | null = null;

    // Private constructor to prevent direct instantiation
    private constructor() {
        this.client = new CosmosClient({
            endpoint: config.cosmos.endpoint,
            key: config.cosmos.key,
            connectionPolicy: {
                requestTimeout: 10000,
            },
            diagnosticLevel: CosmosDbDiagnosticLevel.debug,
        });
    }

    // Public method to get the singleton instance
    public static getInstance(): CosmosDB {
        if (!CosmosDB.instance) {
            CosmosDB.instance = new CosmosDB();
        }
        return CosmosDB.instance;
    }

    // Initialize the database and containers
    public async initialize(): Promise<void> {
        if (this.initialized) {
            return;
        }

        if (this.initializing) {
            return this.initializing;
        }

        this.initializing = this.init();
        return this.initializing;
    }

    // Internal initialization logic
    private async init(): Promise<void> {
        try {
            // Create database if it doesn't exist
            const { database } = await this.client.databases.createIfNotExists({
                id: config.cosmos.database,
                // throughput: 400 // Adjust as needed or make configurable
            });

            // Initialize productsContainer
            this.productsContainer = await this.createContainerIfNotExists(database, 'productsContainer');

            this.initialized = true;
            logger.info(`Database : ${database.id} initialization successfully`);
            logger.info(`Container : ${this.productsContainer.id} initialization successfully`);
        } catch (error: any) {
            if (error instanceof Error) {
                logger.error('Error initializing CosmosDB:', error.message);
                logger.error('Stack Trace:', error.stack);
            } else {
                logger.error('Unexpected error initializing CosmosDB:', JSON.stringify(error, null, 2));
            }
            throw error;
        }
    }

    // Helper method to create a container if it doesn't exist
    private async createContainerIfNotExists(database: any, containerId: string): Promise<Container> {
        const vectorEmbeddingPolicy: VectorEmbeddingPolicy = {
            vectorEmbeddings: [
                {
                    path: "/imageVector",
                    dataType: VectorEmbeddingDataType.Float32,
                    dimensions: 8,
                    distanceFunction: VectorEmbeddingDistanceFunction.DotProduct,
                },
                {
                    path: "/descriptionVector",
                    dataType: VectorEmbeddingDataType.Float32,
                    dimensions: 10,
                    distanceFunction: VectorEmbeddingDistanceFunction.Cosine,
                },
                {
                    path: "/tagsVector",
                    dataType: VectorEmbeddingDataType.Float32,
                    dimensions: 10, // Adjust dimensions as per your embedding model
                    distanceFunction: VectorEmbeddingDistanceFunction.Cosine,
                },
                {
                    path: "/featuresVector",
                    dataType: VectorEmbeddingDataType.Float32,
                    dimensions: 10, // Adjust dimensions as per your embedding model
                    distanceFunction: VectorEmbeddingDistanceFunction.Cosine,
                },
                {
                    path: "/dimensionsVector",
                    dataType: VectorEmbeddingDataType.Float32,
                    dimensions: 4, // weight, width, height, depth
                    distanceFunction: VectorEmbeddingDistanceFunction.Euclidean, // Suitable for numeric vectors
                },
            ],
        };

        const indexingPolicy: IndexingPolicy = {
            vectorIndexes: [
                { path: "/imageVector", type: VectorIndexType.QuantizedFlat },
                { path: "/descriptionVector", type: VectorIndexType.DiskANN },
                { path: "/tagsVector", type: VectorIndexType.DiskANN },
                { path: "/featuresVector", type: VectorIndexType.DiskANN },
                { path: "/dimensionsVector", type: VectorIndexType.DiskANN },
            ],
            includedPaths: [
                {
                    path: "/*",
                },
            ],
            excludedPaths: [
                {
                    path: "/_etag/?"
                },
                {
                    path: "/_rid/?"
                },
                {
                    path: "/_etag/?"
                },
                {
                    path: "/_attachments/?"
                },
                {
                    path: "/_ts/?"
                },
                {
                    path: "/imageVector/*",
                },
                {
                    path: "/descriptionVector/*",
                },
                {
                    path: "/tagsVector/*",
                },
                {
                    path: "/featuresVector/*",
                },
                {
                    path: "/dimensionsVector/*",
                },
            ]
        };

        const { container } = await database.containers.createIfNotExists({
            id: containerId,
            partitionKey: {
                paths: ["/id"],
                version: PartitionKeyDefinitionVersion.V2,
                kind: PartitionKeyKind.Hash,
            },
            vectorEmbeddingPolicy: vectorEmbeddingPolicy,
            indexingPolicy: indexingPolicy,
        });
        return container;
    }


    // Getters for the containers
    private getContainer(container: Container): Container {
        if (!this.initialized) {
            throw new Error('CosmosDB has not been initialized. Call initialize() first.');
        }
        return container;
    }

    public getProductsContainer(): Container {
        return this.getContainer(this.productsContainer);
    }

}

export default CosmosDB;

This service ensures a singleton instance of CosmosDB, initializes the database and containers with the necessary vector policies and indexes.

Defining Product Schema

src/schemas/productSchema.ts:

//src/schemas/productSchema.ts
import { z } from 'zod';

export const productSchema = z.object({
    id: z.string().uuid().optional(),
    name: z.string(),
    brand: z.string(),
    sku: z.string(),
    category: z.string(),
    price: z.number(),
    currency: z.string(),
    stock: z.number(),
    description: z.string(),
    features: z.string(),
    rating: z.number(),
    reviewsCount: z.number(),
    tags: z.array(z.string()),
    imageUrl: z.string().url(),
    manufacturer: z.string(),
    model: z.string(),
    releaseDate: z.string(), 
    warranty: z.string(),
    dimensions: z.object({
        weight: z.string(),
        width: z.string(),
        height: z.string(),
        depth: z.string(),
    }).optional(),
    color: z.string(),
    material: z.string(),
    origin: z.string(),
    descriptionVector: z.array(z.number()).optional(),
    imageVector: z.array(z.number()).optional(),
    tagsVector: z.array(z.number()).optional(),
    featuresVector: z.array(z.number()).optional(),
    dimensionsVector: z.array(z.number()).optional(),
});

export const productIDSchema = z.object({
    id: z.string().uuid(),
});

export const tagsSchema = z.object({
    tags: z.array(z.string()),
});

export const featuresSchema = z.object({
    tags: z.array(z.string()),
});




export type Product = z.infer<typeof productSchema>;
export type ProductID = z.infer<typeof productIDSchema>;
export type Tags = z.infer<typeof tagsSchema>;

Using Zod for schema validation ensures that incoming data adheres to the expected structure.

Configuring Environment Variables

src/config/env.ts:

// src/config/env.ts
import dotenv from 'dotenv';
import { z } from 'zod';

// Load environment variables from .env file
dotenv.config();

// Define the schema for environment variables using zod
const EnvSchema = z.object({
    PORT: z.coerce.number().default(8000), // Set default to 8000
    AZURE_COSMOS_DB_ENDPOINT: z.string({
        required_error: "AZURE_COSMOS_DB_ENDPOINT is required",
        invalid_type_error: "AZURE_COSMOS_DB_ENDPOINT must be a string",
    }),
    AZURE_COSMOS_DB_KEY: z.string({
        required_error: "AZURE_COSMOS_DB_KEY is required",
        invalid_type_error: "AZURE_COSMOS_DB_KEY must be a string",
    }),
    AZURE_COSMOS_DB: z.string({
        required_error: "AZURE_COSMOS_DB is required",
        invalid_type_error: "AZURE_COSMOS_DB must be a string",
    }),
    AZURE_OPENAI_API_KEY: z.string({
        required_error: "AZURE_OPENAI_API_KEY is required",
        invalid_type_error: "AZURE_OPEN_AI_API_KEY must be a string",
    }),
    AZURE_OPENAI_TEXT_EMBEDDING_MODEL_ENDPOINT: z.string({
        required_error: "AZURE_OPENAI_TEXT_EMBEDDING_MODEL_ENDPOINT is required",
        invalid_type_error: "AZURE_OPENAI_TEXT_EMBEDDING_MODEL_ENDPOINT must be a string",
    }),
});

// Parse and validate the environment variables
export const env = EnvSchema.parse(process.env);


// Configuration object consolidating all settings
const config = {
    port: env.PORT,
    cosmos: {
        endpoint: env.AZURE_COSMOS_DB_ENDPOINT,
        key: env.AZURE_COSMOS_DB_KEY,
        database: env.AZURE_COSMOS_DB,
    },
    azureOpenAi: {
        apiKey: env.AZURE_OPENAI_API_KEY,
        models: {
            embedding: {
                endPoint: env.AZURE_OPENAI_TEXT_EMBEDDING_MODEL_ENDPOINT
            },
        },
    },
};


export default config;

This configuration ensures that all necessary environment variables are set and accessible throughout the application.

Setting Up the Express Application

src/app.ts:

//
import express, { Application } from 'express';
import cors from 'cors';
import helmet from 'helmet';
import requestLogger from './middlewares/log';
import productRouter from './routes/productsRoutes';
import { errorHandler } from './middlewares/errorHandler';
import limiter from './middlewares/rateLimiterHandler';

// Create Express server
const app: Application = express();

// CORS options
const corsOptions = {
    origin: 'http://your-allowed-origin.com', // Replace with your allowed origin(s)
    methods: ['GET', 'POST', 'PUT', 'DELETE', 'OPTIONS'],
    allowedHeaders: ['Content-Type'],  // Add headers you want to allow ie 'Authorization'
    // credentials: true, // Enable if you need to pass cookies or authentication tokens
  };

// Middlewares
app.use(cors(corsOptions));  // Enable CORS
app.use(express.json());  // Parse JSON bodies
app.use(express.urlencoded({ extended: true }));  // Parse URL-encoded bodies
app.use(helmet()); // Secure your Express apps by setting various HTTP headers
app.use(requestLogger); // HTTP request logger
app.use(limiter); // Rate limiter middleware to limit(Limit each IP to 100 requests per 1 minutes) 


app.use('/api', productRouter);

app.use(errorHandler); // Error handler middleware( should be the last middleware)

export default app;

This sets up the Express application with necessary middlewares and routes.

Testing the API

Testing dummy data:

  {
    "name": "Wireless Bluetooth Headphones",
    "brand": "SoundMax",
    "sku": "SKU-1001",
    "category": "Electronics",
    "price": 149.99,
    "currency": "USD",
    "stock": 250,
    "description": "Experience unparalleled sound quality with SoundMax Wireless Bluetooth Headphones. Featuring active noise cancellation, 30-hour battery life, and a comfortable over-ear design.",
    "features": "Bluetooth 5.0, Active Noise Cancellation, 30-hour Battery Life, Over-Ear Comfort, Built-in Microphone, Foldable Design",
    "rating": 4.5,
    "reviewsCount": 1200,
    "tags": ["audio", "wireless", "headphones", "electronics", "noise-cancelling"],
    "imageUrl": "https://example.com/images/products/SKU-1001.jpg",
    "manufacturer": "SoundMax Inc.",
    "model": "SM-BT1001",
    "releaseDate": "2023-05-15",
    "warranty": "2 years",
    "dimensions": {
      "weight": "250g",
      "width": "20cm",
      "height": "18cm",
      "depth": "7cm"
    },
    "color": "Black",
    "material": "Plastic and Metal",
    "origin": "USA"
  },
  {
    "name": "Smart LED Light Bulb",
    "brand": "BrightHome",
    "sku": "SKU-1002",
    "category": "Home Automation",
    "price": 19.99,
    "currency": "USD",
    "stock": 500,
    "description": "Illuminate your home with BrightHome Smart LED Light Bulbs. Control brightness and color temperature via smartphone or voice commands.",
    "features": "Wi-Fi Enabled, Voice Control Compatible, Energy Efficient, Adjustable Brightness, Long Lifespan, Easy Installation",
    "rating": 4.3,
    "reviewsCount": 850,
    "tags": ["lighting", "smart home", "LED", "energy-efficient", "voice-control"],
    "imageUrl": "https://example.com/images/products/SKU-1002.jpg",
    "manufacturer": "BrightHome Technologies",
    "model": "BH-SMART-LB",
    "releaseDate": "2023-07-20",
    "warranty": "1 year",
    "dimensions": {
      "weight": "75g",
      "width": "6.5cm",
      "height": "11cm",
      "depth": "6.5cm"
    },
    "color": "White",
    "material": "Glass and Plastic",
    "origin": "Germany"
  },
  {
    "name": "4K Ultra HD Action Camera",
    "brand": "AdventureCam",
    "sku": "SKU-1003",
    "category": "Photography",
    "price": 249.99,
    "currency": "USD",
    "stock": 150,
    "description": "Capture every adventure in stunning 4K Ultra HD with AdventureCam Action Cameras. Waterproof, shockproof, and built for extreme conditions.",
    "features": "4K Video Recording, Waterproof up to 30m, Wi-Fi Connectivity, Image Stabilization, Wide-Angle Lens, Long Battery Life",
    "rating": 4.6,
    "reviewsCount": 620,
    "tags": ["camera", "action", "4K", "waterproof", "photography"],
    "imageUrl": "https://example.com/images/products/SKU-1003.jpg",
    "manufacturer": "AdventureCam LLC",
    "model": "AC-4KUHD",
    "releaseDate": "2023-03-10",
    "warranty": "2 years",
    "dimensions": {
      "weight": "180g",
      "width": "7cm",
      "height": "5cm",
      "depth": "4cm"
    },
    "color": "Red",
    "material": "Aluminum and Plastic",
    "origin": "China"
  },
  {
    "name": "Ergonomic Office Chair",
    "brand": "ComfortSeat",
    "sku": "SKU-1004",
    "category": "Furniture",
    "price": 199.99,
    "currency": "USD",
    "stock": 80,
    "description": "Upgrade your workspace with the ComfortSeat Ergonomic Office Chair. Designed for maximum support and comfort during long working hours.",
    "features": "Adjustable Height, Lumbar Support, Breathable Mesh Back, Reclining Function, Swivel Base, Armrest Adjustment",
    "rating": 4.7,
    "reviewsCount": 430,
    "tags": ["furniture", "office", "ergonomic", "chair", "comfort"],
    "imageUrl": "https://example.com/images/products/SKU-1004.jpg",
    "manufacturer": "ComfortSeat Corp.",
    "model": "CS-ERGONOMIC-01",
    "releaseDate": "2023-01-25",
    "warranty": "3 years",
    "dimensions": {
      "weight": "12kg",
      "width": "70cm",
      "height": "120cm",
      "depth": "70cm"
    },
    "color": "Gray",
    "material": "Mesh and Metal",
    "origin": "Sweden"
  },
  {
    "name": "Stainless Steel Blender",
    "brand": "KitchenPro",
    "sku": "SKU-1005",
    "category": "Kitchen Appliances",
    "price": 89.99,
    "currency": "USD",
    "stock": 300,
    "description": "Blend smoothies, soups, and more with the KitchenPro Stainless Steel Blender. Powerful motor and durable blades for all your blending needs.",
    "features": "1000W Motor, 1.5L Capacity, Stainless Steel Blades, Multiple Speed Settings, Pulse Function, Easy Clean",
    "rating": 4.4,
    "reviewsCount": 920,
    "tags": ["kitchen", "blender", "appliances", "stainless steel", "smoothies"],
    "imageUrl": "https://example.com/images/products/SKU-1005.jpg",
    "manufacturer": "KitchenPro Industries",
    "model": "KP-BLEND-1000",
    "releaseDate": "2023-04-18",
    "warranty": "2 years",
    "dimensions": {
      "weight": "3kg",
      "width": "15cm",
      "height": "40cm",
      "depth": "15cm"
    },
    "color": "Silver",
    "material": "Stainless Steel and Plastic",
    "origin": "Italy"
  },
  {
    "name": "Fitness Tracker Watch",
    "brand": "HealthMate",
    "sku": "SKU-1006",
    "category": "Wearables",
    "price": 59.99,
    "currency": "USD",
    "stock": 400,
    "description": "Monitor your health and fitness goals with the HealthMate Fitness Tracker Watch. Features heart rate monitoring, step counting, and sleep tracking.",
    "features": "Heart Rate Monitor, Step Counter, Sleep Tracking, Waterproof, Smartphone Notifications, Long Battery Life",
    "rating": 4.2,
    "reviewsCount": 1500,
    "tags": ["fitness", "tracker", "watch", "wearables", "health"],
    "imageUrl": "https://example.com/images/products/SKU-1006.jpg",
    "manufacturer": "HealthMate LLC",
    "model": "HM-FIT-2023",
    "releaseDate": "2023-02-05",
    "warranty": "1 year",
    "dimensions": {
      "weight": "50g",
      "width": "4cm",
      "height": "4cm",
      "depth": "1cm"
    },
    "color": "Blue",
    "material": "Silicone and Plastic",
    "origin": "South Korea"
  },
  {
    "name": "Portable Power Bank 20000mAh",
    "brand": "ChargeUp",
    "sku": "SKU-1007",
    "category": "Mobile Accessories",
    "price": 39.99,
    "currency": "USD",
    "stock": 600,
    "description": "Stay powered on the go with the ChargeUp Portable Power Bank. High-capacity 20000mAh battery ensures your devices stay charged throughout the day.",
    "features": "20000mAh Capacity, Dual USB Ports, Fast Charging, LED Indicator, Lightweight Design, Durable Build",
    "rating": 4.5,
    "reviewsCount": 2100,
    "tags": ["power bank", "portable", "charging", "mobile", "accessories"],
    "imageUrl": "https://example.com/images/products/SKU-1007.jpg",
    "manufacturer": "ChargeUp Technologies",
    "model": "CU-PB20000",
    "releaseDate": "2023-06-12",
    "warranty": "1 year",
    "dimensions": {
      "weight": "500g",
      "width": "7cm",
      "height": "18cm",
      "depth": "7cm"
    },
    "color": "Black",
    "material": "Aluminum and Plastic",
    "origin": "Japan"
  },
  {
    "name": "Smart Thermostat",
    "brand": "HomeSmart",
    "sku": "SKU-1008",
    "category": "Home Automation",
    "price": 129.99,
    "currency": "USD",
    "stock": 180,
    "description": "Optimize your home's temperature with the HomeSmart Smart Thermostat. Features remote control, scheduling, and energy usage reports.",
    "features": "Remote Control via App, Scheduling, Energy Usage Reports, Voice Control Compatible, Easy Installation, Sleek Design",
    "rating": 4.6,
    "reviewsCount": 760,
    "tags": ["thermostat", "smart home", "energy-efficient", "automation", "control"],
    "imageUrl": "https://example.com/images/products/SKU-1008.jpg",
    "manufacturer": "HomeSmart Inc.",
    "model": "HS-THERMO-2023",
    "releaseDate": "2023-08-01",
    "warranty": "2 years",
    "dimensions": {
      "weight": "300g",
      "width": "10cm",
      "height": "15cm",
      "depth": "3cm"
    },
    "color": "White",
    "material": "Plastic and Metal",
    "origin": "USA"
  },
  {
    "name": "Gaming Mechanical Keyboard",
    "brand": "GameMaster",
    "sku": "SKU-1009",
    "category": "Computer Accessories",
    "price": 109.99,
    "currency": "USD",
    "stock": 220,
    "description": "Enhance your gaming experience with the GameMaster Mechanical Keyboard. Features customizable RGB lighting, tactile switches, and durable build.",
    "features": "Mechanical Switches, RGB Backlighting, Programmable Keys, N-Key Rollover, USB Passthrough, Detachable Wrist Rest",
    "rating": 4.8,
    "reviewsCount": 980,
    "tags": ["keyboard", "gaming", "mechanical", "RGB", "computer"],
    "imageUrl": "https://example.com/images/products/SKU-1009.jpg",
    "manufacturer": "GameMaster Technologies",
    "model": "GM-MK1009",
    "releaseDate": "2023-09-10",
    "warranty": "3 years",
    "dimensions": {
      "weight": "900g",
      "width": "45cm",
      "height": "5cm",
      "depth": "15cm"
    },
    "color": "Black",
    "material": "Aluminum and Plastic",
    "origin": "Taiwan"
  },
  {
    "name": "Wireless Charging Pad",
    "brand": "ChargeWave",
    "sku": "SKU-1010",
    "category": "Mobile Accessories",
    "price": 29.99,
    "currency": "USD",
    "stock": 450,
    "description": "Conveniently charge your devices with the ChargeWave Wireless Charging Pad. Compatible with all Qi-enabled devices and features fast charging capabilities.",
    "features": "Qi-Compatible, Fast Charging, Non-Slip Surface, LED Indicator, Slim Design, Overcharge Protection",
    "rating": 4.3,
    "reviewsCount": 1340,
    "tags": ["charging", "wireless", "mobile", "accessories", "fast-charging"],
    "imageUrl": "https://example.com/images/products/SKU-1010.jpg",
    "manufacturer": "ChargeWave Ltd.",
    "model": "CW-WCP300",
    "releaseDate": "2023-07-05",
    "warranty": "1 year",
    "dimensions": {
      "weight": "200g",
      "width": "10cm",
      "height": "1cm",
      "depth": "10cm"
    },
    "color": "White",
    "material": "Plastic",
    "origin": "South Korea"
  }

With the API fully set up, let’s explore how to interact with it using various endpoints.

Creating a Product

Endpoint: POST http://localhost:8000/api/products

Headers: 

Content-Type: application/json

Request Body:

{
  "name": "Smart LED Light Bulb",
  "brand": "BrightHome",
  "sku": "SKU-1002",
  "category": "Home Automation",
  "price": 19.99,
  "currency": "USD",
  "stock": 500,
  "description": "Illuminate your home with BrightHome Smart LED Light Bulbs. Control brightness and color temperature via smartphone or voice commands.",
  "features": "Wi-Fi Enabled, Voice Control Compatible, Energy Efficient, Adjustable Brightness, Long Lifespan, Easy Installation",
  "rating": 4.3,
  "reviewsCount": 850,
  "tags": [
    "lighting",
    "smart home",
    "LED",
    "energy-efficient",
    "voice-control"
  ],
  "imageUrl": "https://example.com/images/products/SKU-1002.jpg",
  "manufacturer": "BrightHome Technologies",
  "model": "BH-SMART-LB",
  "releaseDate": "2023-07-20",
  "warranty": "1 year",
  "dimensions": {
    "weight": "75g",
    "width": "6.5cm",
    "height": "11cm",
    "depth": "6.5cm"
  },
  "color": "White",
  "material": "Glass and Plastic",
  "origin": "Germany"
}

Response:

{
  "status": "success",
  "data": {
    "id": "a2555d74-686a-4c34-af2b-c63fa2b9ed2f",
    "name": "Smart LED Light Bulb",
    "brand": "BrightHome",
    "sku": "SKU-1002",
    "category": "Home Automation",
    "price": 19.99,
    "currency": "USD",
    "stock": 500,
    "description": "Illuminate your home with BrightHome Smart LED Light Bulbs. Control brightness and color temperature via smartphone or voice commands.",
    "features": "Wi-Fi Enabled, Voice Control Compatible, Energy Efficient, Adjustable Brightness, Long Lifespan, Easy Installation",
    "rating": 4.3,
    "reviewsCount": 850,
    "tags": [
      "lighting",
      "smart home",
      "LED",
      "energy-efficient",
      "voice-control"
    ],
    "imageUrl": "https://example.com/images/products/SKU-1002.jpg",
    "manufacturer": "BrightHome Technologies",
    "model": "BH-SMART-LB",
    "releaseDate": "2023-07-20",
    "warranty": "1 year",
    "dimensions": {
      "weight": "75g",
      "width": "6.5cm",
      "height": "11cm",
      "depth": "6.5cm"
    },
    "color": "White",
    "material": "Glass and Plastic",
    "origin": "Germany",
    "descriptionVector": [/* generated vector */],
    "tagsVector": [/* generated vector */],
    "featuresVector": [/* generated vector */]
  }
}

Upon creation, the API generates vector embeddings for the description, tags, and features, storing them alongside the product data.

Retrieving Products

Endpoint: GET http://localhost:8000/api/products

Headers:

Accept: application/json

Response:

{
  "status": "success",
  "data": [
    {
      "id": "a2555d74-686a-4c34-af2b-c63fa2b9ed2f",
      "name": "Smart LED Light Bulb",
      "brand": "BrightHome",
      "sku": "SKU-1002",
      "category": "Home Automation",
      "price": 19.99,
      "currency": "USD",
      "stock": 500,
      "description": "Illuminate your home with BrightHome Smart LED Light Bulbs. Control brightness and color temperature via smartphone or voice commands.",
      "features": "Wi-Fi Enabled, Voice Control Compatible, Energy Efficient, Adjustable Brightness, Long Lifespan, Easy Installation",
      "rating": 4.3,
      "reviewsCount": 850,
      "tags": [
        "lighting",
        "smart home",
        "LED",
        "energy-efficient",
        "voice-control"
      ],
      "imageUrl": "https://example.com/images/products/SKU-1002.jpg",
      "manufacturer": "BrightHome Technologies",
      "model": "BH-SMART-LB",
      "releaseDate": "2023-07-20",
      "warranty": "1 year",
      "dimensions": {
        "weight": "75g",
        "width": "6.5cm",
        "height": "11cm",
        "depth": "6.5cm"
      },
      "color": "White",
      "material": "Glass and Plastic",
      "origin": "Germany",
      "descriptionVector": [/* vector */],
      "tagsVector": [/* vector */],
      "featuresVector": [/* vector */]
    },
    // More products...
  ]
}

Performing Vector Similarity Searches

Vector similarity searches allow users to find products that are semantically similar to their queries. Below are examples of how to perform such searches.

Search by Tags Vector

Endpoint: POST http://localhost:8000/api/products/search/tags

Headers:

Content-Type: application/json

Request Body:

{
  "queryVector": ["lighting", "smart home"],
  "top": 5
}

Response:

{
  "status": "success",
  "data": [
    {
      "id": "a2555d74-686a-4c34-af2b-c63fa2b9ed2f",
      "name": "Smart LED Light Bulb",
      "brand": "BrightHome",
      "sku": "SKU-1002",
      // Other product details...
      "SimilarityScore": 0.95
    },
    // Top 5 similar products...
  ]
}

Search by Features Vector

Endpoint: POST http://localhost:8000/api/products/search/features

Headers:

Content-Type: application/json

Request Body:

{
  "queryVector": ["Wi-Fi Enabled", "Voice Control"],
  "top": 2
}

Response:

{
  "status": "success",
  "data": [
    {
      "id": "a2555d74-686a-4c34-af2b-c63fa2b9ed2f",
      "name": "Smart LED Light Bulb",
      "brand": "BrightHome",
      "sku": "SKU-1002",
      // Other product details...
      "SimilarityScore": 0.92
    },
    // Top 2 similar products...
  ]
}

Search by Description Vector

Endpoint: POST http://localhost:8000/api/products/search/description

Headers:

Content-Type: application/json

Request Body:

{
  "queryVector": ["energy efficient smart lighting"],
  "top": 10
}

Response:

{
  "status": "success",
  "data": [
    {
      "id": "a2555d74-686a-4c34-af2b-c63fa2b9ed2f",
      "name": "Smart LED Light Bulb",
      "brand": "BrightHome",
      "sku": "SKU-1002",
      // Other product details...
      "SimilarityScore": 0.98
    },
    // Top 10 similar products...
  ]
}

These searches utilize the vector embeddings to retrieve products that closely match the semantic intent of the query, providing more relevant results compared to traditional keyword searches.

Conclusion

Integrating vector similarity search into your e-commerce API using Azure Cosmos DB enhances the search functionality by understanding the semantic relationships between products and user queries. This leads to improved user satisfaction, higher engagement, and increased sales. By leveraging Azure’s robust infrastructure and advanced embedding techniques, you can stay ahead in the competitive e-commerce landscape.

Key Takeaways:

• Vector Embeddings: Transforming textual and categorical data into high-dimensional vectors enables nuanced similarity comparisons.
• Azure Cosmos DB Vector Search: Offers efficient indexing and querying capabilities tailored for vector data.
• API Integration: Seamlessly integrating vector search into your API enhances the overall search experience.
• Scalability: Azure Cosmos DB ensures that your vector search capabilities scale with your growing product catalog.

Embark on implementing vector similarity search in your e-commerce platform to deliver smarter and more intuitive search experiences for your users.

Read More 

• What is Azure OpenAI Service?
• Azure OpenAI Service models
• Tutorial: Explore Azure OpenAI Service embeddings and document search
• Index and query vectors in Azure Cosmos DB for NoSQL in JavaScript
• What is a vector database
• Vector database in Azure Cosmos DB for NoSQL
• Vector database in Azure Cosmos DB for MongoDB
• Announcing JavaScript SDK v4 for Azure Cosmos DB

​Microsoft Tech Community – Latest Blogs –Read More 

SQL Server 采用客户端 / 服务器架构，客户端应用程序通过网络连接与 SQL Server 服务器进行通信。客户端可以是各种类型的应用程序，如桌面应用程序、Web 应用程序、移动应用程序等，服务器端负责接收客户端的请求，执行相应的数据库操作，并将结果返回给客户端。服务器端由多个组件组成，包括 SQL Server 引擎、存储引擎、网络协议栈等。

​SQL Server 采用客户端 / 服务器架构，客户端应用程序通过网络连接与 SQL Server 服务器进行通信。客户端可以是各种类型的应用程序，如桌面应用程序、Web 应用程序、移动应用程序等，服务器端负责接收客户端的请求，执行相应的数据库操作，并将结果返回给客户端。服务器端由多个组件组成，包括 SQL Server 引擎、存储引擎、网络协议栈等。  Read More

​

Hello everyone,

I got a new iPhone 15 Pro Max and I’m trying to find a reliable way to mirror my iPhone’s screen onto my PC running Windows 11. I’ve seen several methods online, but I’m not sure which ones are the most effective and user-friendly. Ideally, I’m looking for a solution that offers high-quality mirroring without too much lag, and it would be great if it also supports audio transmission.

Has anyone here successfully mirrored their iPhone to a Windows 11 PC? I’m open to both free and paid solutions, as long as they are reliable and straightforward to use. Thank you in advance for your suggestions!

​Hello everyone,I got a new iPhone 15 Pro Max and I’m trying to find a reliable way to mirror my iPhone’s screen onto my PC running Windows 11. I’ve seen several methods online, but I’m not sure which ones are the most effective and user-friendly. Ideally, I’m looking for a solution that offers high-quality mirroring without too much lag, and it would be great if it also supports audio transmission. Has anyone here successfully mirrored their iPhone to a Windows 11 PC? I’m open to both free and paid solutions, as long as they are reliable and straightforward to use. Thank you in advance for your suggestions!  Read More

​